195704

April

1957

Division of

labor

in

Scientific Digital

Computer

Service Facilities · . .

l.

N.

Caplan

Cam

Profile Design With The Univac 120 · . .

louis

D.

Grey

Supplementing

Electronic Equipment With a

Modern

Communications System

Use of a

Computer

for

Certain

Operations

of

Classification

Monroe

M.

Koontz

Andrew

D.

Booth

Robots

and

Automata:

A Short History -Bibliography

·

..

J.

T.

Culbertson

New

Products

and

Ideas

Western

Joint

Computer

Conference,

los

Angeles,

February 26-28, 1957 -

Program,

Titles,

and

Abstracts

-

~

an

open

letter

to

The

General Electric Company has announced

the

establishment of

new facilities in Phoenix, Arizona for the development

and

manufac-

ture

of a full line of digital

and

analog, special afld general purpose

computers

and

peripheral machines.

It

is intended to apply this equip.

ment to a broader range of commercial

and

industrial problems than

has

been attempted previously by any single 'company.

Company management is supporting this objective by building

up

a

strong activity

in

the

general

area

of computer programming. Pro-

posed machines will be simulated on currently available equipment,

research in programming methods will be undertaken,

and

investiga·

tion will be sponsored in information handling fields beyond engi·

neering calculation

and

business

data

processing.

Programming will be done to

support

the application of ERMA,

and

other equipment under construction

and

in

the

logical design stage,

to customer problems.

The

Applications Section is operating a large IBM 704 installation,

including

peripheral

tape

equipment

and

auxiliary

punch

card

machines. Analog and digital equipment of the most advanced sort

will be added as available,

and

training in its application will be

offered both within the Department

and

in

cooperation with the

appropriate schools of Arizona State College.

Phone collect WOodland 7·2001 f9r further details

or

send your reply

in strict confidence.

*,7(t:o

S

Dar.q

J.

GaoscH

M~;

-Applications Section

COMPUTERS

AND

CYBERNETICS

•

ROBOTS

•

AUTOMATION

AUTOMATIC

CONTROL

Vol.

6,

No.

4

April,

1957

ESTABLISHED

SEPTEMBER,

1951

ARTICLES

AND

PAPERS

Division

of

Labor

in

Scientific

Digital

Computer

Service

Facilities

Cam

Profile

Design

With

the

Univac

120

Supplementing

Electronic

Equipment

With a

Modern

Communications

System

Use

of

a

Computer

for

Certain

Operations

of

Classification

Robots

and

Automata: A

Short

History

-

Bibliography

(Part

2)

NEW PRODUCTS

AND

IDEAS

REFERENCE

INFORMATION

L.

N.

Caplan

Louis

D.

Grey

Monroe

M. Koontz

Andrew

D. Booth

James

T.

Culbertson

Western

Joint

Computer

Conference,

Los

Angeles,

February

26-28,

1957,

Program,

Titles,

and

Abstracts

Association

for

Computing

Machinery,

Los

Angeles

Chapter,

Meeting,

Los

Angeles,

March

1,

1957,

"New

Computers:

A

Report

from

the

Manufacturer

s",

Progr

am

Symposium

on

Systems

for

Information

Retrieval,

Western

Reserve

University

School of

Library

Science,

Cleveland,

Ohio,

April

15-17,

1957,

Program

FORUM

On

Glossaries

and

Malapropisms

The

Editor's

Notes

Index

of

Notices

Advertising

Index

•..

Alston

S.

Householder

Editor:

Edmund

C.

Berkeley

Assistant

Edi

tors:

Neil

D.

Macdonald.,

F.

L. Walker

Contributing

Editors:

Andrew

D Booth, John

W.

Carr,III,

Alston

S. Householder

Advisory Committee: Samuel

B.

Williams,

Herbert

F.

Mitchell

Jr.,

Howard

T. Engstrom,

Alston

S. Householder,

H.

Jefferson

Mills,

Jr.

Publisher:

Berkeley

Enterprises,

Inc.

815

Washington

Street,

Newtonville

60, Mass. -

Decatur

2-5453

or

2-3928

Advertising

Representatives:

New

York -

Milton

L.

Kaye,

601

Madison Ave.,

New

York 22, N.Y.,

Plaza

5-4680

San

Francisco

-W.A.Babcock,

60S

Market

St.,

San

Francisco

5,

Calif.,

Yukon

2-3954

6

10

12

18

20

8

25

30

31

33

6

46

Los Angeles -Wentworth

F.

Green,

439

So. Western Ave., Los Angeles

5,

Calif.,

Dunki

rk

7

-8135

f:l

sewhere -

the

Pub1.isher

COMPUTERS

AND

AUlOMATIONis

published

monthly. Copyright,

195~

by

Berkeley

Enterprises,

Inc.

Subsc'ription

rate!!:

in

the

Uni

ted

States

-one

year

S5.50,

two

years

$10.50;

in

Canada -one

year

$6.00,

two

years

$11.00;

elsewhere -one

year

S6.5e,

two

years

$12.50

~tered

as· second

class

matter

at~_he~P_o_st~O_ff_i_c_e~,_N_e_w_~_o_r~k,~N_.Y_.~~~~~~~~~~~~

- 3 -

TH·E EDITOR'S NOTES

NEW PRODUCTS AND IDEAS

We

take

pleasure

in

putting

into

this

issue

of

"Computers

and

Automation"

brief

accounts

of

two

important

new

developments

in

the

computer

field,

neither

yet

completed.

They

show

again

how

un-

limited

this

field

is.

One

is

an

accwnt

of

Project

STRETCH,

in

which

the

U.S.

Atomic

Energy

Com-

mission

and

International

Business

Machines

Corp.

are

engaged.

This

machine

will

gulp a

million

pro-

gram

instructions

a

second.

The

other

is

an

ac-

COlmt

of

the

Datamatic

1000,

which

is

due

to

be

complete

in

the

autumn.

It

will

sort

or

merge

5000

items

of

12

decimal

digits

per

second.

Both

set

new

marks.

We

tried

very

hard

to

get

a

third

fron-

tier

computer

story,

which

has

been

described

at

recent

computer

meetings;

we

wrote

letters

and

wired,

but

received

no

response

to

any of

our

re-

quests.

To

your

editor,

the

computer

field

is

still

extraordinary.

He

started

work

upon

graduating

from

college

in

1930

in

the

a9tuarial

department

of

a

life

insurance

company

as

an

actuarial

clerk.

For

four

years,

his

equipment

consisted

of

large

sheets

of

ruled

paper

for

entering

figures,

tables

of

actuarial

functions

and

formulas,

and

desk

cal-

culating

machines,

where

to

shift

from

one

figure

column

to

the

next,

you

turned

a

handle.

In

the

13

years

since

1944,

the

change

has

been

tremend-

ous.

To

mathematics

with

wings

have

been

added

arithmetic

and

calculation

with

wings.

It

is

a

satisfaction

to

think

of the

important

new

products

and

ideas

in

the

computer

field

which

without

any

doubt

will

come

into

existence

in

the

to

100

years:

-

Machines

that

will

recognize

-

Machines

that

will

play

chess

or

other

games

at

least

as

well

as

nine

tenths

of

human

players

-

Machines

that

will

handle

1&t'1guage

as

intell~gently

as

sixty

percent

of

hu-

man

beings

-

Machines

that

will

teach,supplement-

ing

and

aiding

the

work

of

teachers

in

school

-

Machines

that

will

treat

over

ninety

percent

of

cases

of

mental

illness,

and

diagnose

nearly

a

hundred

per-

cent

of

all

cases

of

illness

-

Machines

that

will

calculate

alternat-

ive

solutions

to

social

and

economic

problems,

and

al

ternative

plans

for

the

production

of a whole

society

And

more

besides.

There

is

no

theoretical

barr-

ier

to

any

of

these

machines.

- 4 -

WHO'S WHO

IN

THE COMPUTER

FIELD,

1956-57

This

extra

number

of

"Computers

and

Auto-

mation",

212

pages

long,

went

to

press

on

March

11.

It

contains

199

pages

of

entries,

with

about

61

entries

per

page,

each

showing

name,

address,

and

some

information

for

about

12,000

computer

people.

Please

look

at

your

entry

in

the

directory,

and

please

tell

us

of

any

revision

needed.

We

ex-

pect

to

print

revisions

in

early

issues

of

"Comput-

ers

and

Automation".

THE COMPUTER DIRECTORY

AND

BUYERS' GUIDE,

1957

This

regular

number

of

"Computers

and

Auto-

mation",

the

June

issue,

will

contain

as

Part

1,

a

"Roster

of

Organizations

in

the

Computer

Field",

and

as

Part

2,

"Buyers'

Guide to

the

Computer

Field:

Products

and

Services

for

Sale

or

Rent".

For

roster

entry

forms,

see

page

43

of

the

March

issue.

The

closing

date

for

information

to

be

pub-

lished

in

the

directory

will

be

about

April

30.

*----------

*

----------*

NEW

PRODUCTS

AND

IDEAS

1.

The

Ds.tamatic

1000

Computer

2.

The

Project

STRETCH

Computer

(IBM)

3.

Programming

With Soap (Symbolic

Op-

timum

Assembly

Programming)

~

8

9

* * *

INDEX

OF

NOTICES

For

Information

on:

See

Page:

Adve

rtising

Index

Advertising

Rates

and

Specifications

Back

Copies

Bulk

Subscription

Rates

Manuscripts

Reader's

Inquiry

Form

Special

Issues

37,

see

Mar.

see

Mar.

46

44

40

32

issue

46

issue

Address

Changes:

If

your

address

changes,

please

send

us

both

your

new

and

your

old

address

(torn

off

from

the

wrapper

if

possible),

and

all~w

three

weeks

for

the

change.

Microton hardness test typifies

quality

control measures that ieat'e nothing

to

chance at

Automatic

Electric

STANDARDS

THAT

DETERMINE

RELAY

QUALlT!

the

ideal

relay

iron

Only

soft,

pure

iron

assures

proper

release,

even

after

millions

of

operations.

In

service,

many

relays

get

progres-

sively

slower

to

release,

until

finally

the

armature

hangs

up

permanently.

Excessively

"hard"

magnetic

relay

iron

often

is

to

blame.

But

in

all

Automatic

Electric

relays,

the

mag-

netic

iron

is

so

soft

and

pure

that

it

saturates

quickly,

yet

the

flux dies

out

instantly.

Even

after

millions

of

operations!

This

improved

relay

iron

is

made

and

rolled

to

our

own

exacting

speci-

AUTOMATIC

fica

tions.

Chemical

analysis

then

makes

certain

that

no

magnetic

ca-

pabilities

have

been

lost.

Annealing

is rigidly

controlled,

and

grain

size

and

temper

carefully

checked.

Elon-

gation,

hardness,

permeability,

and

density

tests

further

safeguard

this

pampered

material.

Result:

not

the

ideal

relay

iron,

but

the most nearly

perfect iron available.

This

is

no

isolated

example.

Ex-

haustive

tests

prove

that

all

our

raw

material

is

equally

fine.

It's

one

of

the

many

reasons

why

Automatic

Electric

relays

enjoy

far

longer

life.

ELECTRIC

....................................

·

...................................

Series

SQPC

Relay

lor

printed

circuitry

applications.

Write

today for

Bulletin

RH-9.

Automatic

Electric Sales Corpora-

tion, Chicago 7.

In

Canada:

Automatic

Electric Sales (Canada)

Ltd.,

Toronto.

Offices

in

principal

cities.

DIVISION

OF LABOR IN SCIENTIFIC

DIGITAL COMPUTER SERVICE FACILITIES

L.N. Caplan

463

Irving

St.

Dayton 9,Ohio

The

past

five

years

have seen a

'~opsy

like"

growth

of

digital

computer

scientific

service

facilities

in

American

industry.

Since

this

growth has been

so

rapid,

labor

division

has been done

on

an

expediency

basis

rather

than

by

planned

organizational

proced-

ure.

The

result

is

an

overlapping

of

techni-

cal

and

semi-technical

skills

in

particular

jobs

within

computer

service

organizations.

With

the

shortage

of

highly

skilled

personnel

that

exists

at

the

present

time,

efficient

use

of

personnel

is

a

serious

problem.

Not

much

has been

written

about

this

par-

ticular

aspect

of

computer

facility

organiza-

tion.

The

following

theory

of

organization

is

a

possible

guide toward a

solution

of

this

problem.

An

endeavor

will

be

made

to

delineate

the

types

of

work

to

be done

within

the

organ-

ization,

to

estimate

skills

required

for

the

work, and

to

assign

specific

job

classifica-

tions.

Thus,

the

company

thinking

of adding

a

digital

computer

facility

will

get

at

least

a framework

idea

of

what

is

required

in

the

way

of

organization.

Existing

facilities

may

gain

an

idea

or

so

for

making

better

use of

the

personnel

they

now

possess.

Problems brought

to

the computing

facil-

ity

can

usually

be

classified

into

three

types.

The

division

of

labor

will

be

made

upon

the

basis

of

solution

of

these

three

types

--

that

is,

the

people

called

upon

to

solve

problems

of

a

certain

type

will

be

placed

in

a

specific

job

classification.

It

should

be

noted

that

solution

of

any problem

involves

cooperation

and exchange

of

information

between

job

class-

ifications.

The

work should be arranged

so

that

employees

of

one

job

classification

learn

from

the

employees

of

another.

There

is

noth-

ing

so

conducive

to

labor

turnover

as

assign-

ing

a

specific

function,

such as machine

cod-

ing

only,

to

members

of

the

organization.

Job

Classification

for

Problems

of

Type

I

In

all

mathematical and

engineering

prob-

-6-

lems a

good

portion

of

the

mathematics involved

is

simple

algebraic

and

transcendental

equation

solution.

Therefore,

the

solution

of

these

equations

is

the

first

type problem

to

be

con-

sidered.

If

a range

of

variables

is

desired

in

combination

with

a

large

number

of

values

for

the

coefficients

of

the

equation,

hand

calculation

(although

possible)

is

now

prohib-

itive

time

wise;

and

the

solutions

of

these

equations

are

machine computation problems.

The

skills

required

for

the

solution

of

this

type

of problem are: A working knowledge

of

algebra

and

trigonometry,

coding knowledge

of

the

machine

being

used, and a

clever

knack

for

machine

logic.

In

solving

these

problems

machine

errors,

errors

in

logic,

and

other

simple numerical mistakes

are

invariably

en-

countered.

Therefore,

in

addition

to

the

skills

listed

above,

the

person

solving

the

problem must

be

able

to

diagnose

trouble

in

the

case

of

incorrect

answers

or

failure

of

the

problem

in

the

machine. This

calls

for

an

exacting

knowledge

of

the

machine

itself,

as

well

as

an

analytic

mind.

The

two

latter

features

are

no

small

part

of

the

job.

Let us

call

the

job

classification

for

solving

this

type

of

problem, "Programmer".

Since

even

the

latest

computing machines

can only add,

subtract,

multiply,

and

divide,

if

the

equations

given

to

the

programmer con-

tain

functions

of

a

variable,

such as exponen-

tial

of

X,

root

of

X,

sin

X,

etc.,

the

pro-

grammer

must have

available

methods

of

using

add,

subtract,

divide,

and

multiply

to

compute

these

functions.

These methods

are

commonly

known

in

the

trade

as

sub-routines.

We

shall

leave

the

designing

of

the

sub-routines

to

the

second

job

classification.

Job

Classification

for

Problems

of

Type

II

Under problems

of

Type

II

consider:

Solutions

of

differential-integral

equa-

tions

with

variable

coefficients

that

have

no

known

or

simple

closed

form

solution.

Division

of

Labor

Systems of

algebraic

or

differential

equations

with

variable

coefficients.

Functions as log x,

sin

x, x

lIn.

eX.

These problems

all

have

one

thing

in

common

when

done

on

a computer.

An

approximating

series

or

iterative

procedure must

be

used.

The

solution

of

these

problems

requires

all

of the

skills

embodied in

solving

problems

of

Type

I,plus

a working knowledge of numer-

ical

analysis

techniques.

Consider

evaluat-

ing

an

integral.

Will

Simpson's Rule

for

solution

furnish

accurate

enough

~nswers?

Perhaps

it

is

necessary

to

start

the

solution

with

one

method

and

later

change tG

another.

Not only

is

there

a

difference

of accuracy

in

these

methods, but

there

may

be

a

large

amount

of

machine time

difference.

Since

machine time

for

the

larger

machines

is

estimated

at

costing

between 4

and

12

dollars

per minute, the

method

will

make

quite

a

dif-

ference

in

the

cost

of

solving

large

problems.

The

better

the problem

solver

is

acquainted

with the machine

itself,

the

better

he

is

able

to

determine what numerical techniques

to

use

with

respect

to

accuracy of

solution,

economy

of time,

and

trouble

shooting

pro-

cedure

in

case

of

difficulties

arising

in

machine

solution.

Type

II

problem

solvers

are

"Numerical

Analysts".

As

mentioned

previously,

the ana-

lyst

must supply machine

subroutines

for

eval-

uating

mathematical

functions

to

the Pro-

grammer.

The

Programmer

is

then

able

to

use

these

routines

in

solving

equations

involv-

ing

these

functions.

Once

the numerical an-

alysis

has been done

on

a problem (which

means

it

has been reduced

to

algebraic'

equations)

the

problem

may

be

turned over

to

the

pro-

grammer.

Job

Classification

for

Problems

of

Type

III

Type

III

problems

deal

with the design

of

m~thematical

models of

physical

systems,

such as chemical

processes,

rockets,

jet

and

airframe

performance,

and

electronic

servo-

mechanisms.

The

mathematical

model

must

be

suitable

for

machine computation. Such a

system

may

save

building

pilot

models,

since

the

experimentation

is

done mathematically

by

varying design parameters numerically

in

the

machine. In

some

instances,

control

over

the

actual

processes

is

desired

through the

use of a

stored

program

digital

computer*.

*See

Fortune,

April

21, 1956.

The

article

~ntitled

"Automation"

furnishes

informa-

tion

about

this

type

of

problem.

- 7 -

Type

III

problems

differ

from Types I

and

II

in

that

in

the

latter

problems the equations

were

furnished

by

the person

desiring

solution.

It

is

assumed

that

the equations

arose

from

problems

that

originated

in

Research

and

De-

velopment. In

this

type problem the

solver

must

help

formulate the

equations.

His

de-

cisions

are

based

upon

his

knowledge

of

what

the

machine can do.

One

might

think

that

the

physical

system determines the formulation

of

the mathematical model, but

this

is

not

true.

Consider a problem where maximization

of

a

quantity

is

desired.

It

is

the

problem solver's

decision

as

to

whether

linear

programming or

Lagrange

mUltiplication

method

is

most appro-

priate.

The

design of

physical

systems

is

usually

done

by

an

engineering group. Such a

group might include a chemical,

electrical,

mechanical

engineer,

and

perhaps a

physicist.

If

the problem

is

to

be

solved

via

mathemati-

cal

machine models, the group

will

have

to

include a person from the computer

section.

This person can

be

called

a "Computer Appli-

cations

Consultant".

In

order

to

design a

mathematical

model

for

a

physical

system, the

problem

solver

must

be

able

to

understand

and

communicate with the engineers working

on

the

project.

Therefore,

a broad

engineering

back-

ground

is

desired.

In

addition

to

having

all

the

knowledge

required

to

solve

problems of

Type

I

and

II,

the

consultant

must have

an

excellent

knowledge of mathematical techniques,

especially

of

recent

developments.

He

should

know

which

portions

of the problem

are

best

solved

by

analog machines

rather

than

digital

computers.

Personality

will

playa

large

part

in

his

job,as

the

consultant

must

be

a team

worker

and

be

able

to

consult

harmoniously

with the

engineering

group.

After

the

model

has been formed

by

the

consultant,

he

turns

this

model

over

to

the

numerical

analyst.

The

analyst

having reform-

ulated

the

model

into

algebraic

equations

turns

the

problem over

to

the programmers

for

coding.

After

the

problem has been completely

solved,

the

consultant

must

help

interpret

the

results

for

the

engineering

group.

There

is

one

other

position

that

must

be

included

in

the

organizational

setup.

This

is

a "Machine Techniques

Specialist"

to

prepare

formats

for

machine

input

and

output,

design

input

and

output methods, keep

abreast

of

new

techniques

that

may

be

used

on

the

machine,

and

teach

the

how

and

what

required

to

use

the

machine

for

problem

solution.

The

job

calls

for

a person

who

has

grown

up

wi

th

com-

puting

machinery

in

the

past

five

years

rather

than a mathematician. Often

the

company

sup-

plying

the

machine

will

furnish

a machine

techniques

specialist

for

the

first

few

months

of

installation.

(cont'd

on page 36)

NEW

PRODUCTS

AND

IDEAS

THE DATAMATIC 1000

CO:Mi?UTER

Henry

W.

Schrimpf

The

Datamatic

Corporation

Newton Highlands,

Mass.

The

first

one of the

Datamatic

19Q,0

computers

made

by

t11e

Datamatic

Corporation,

formed

1955

and jointly owned

by

Minneapolis-Honeywell

Regu-

lator

Co. and Raytheon Mfg.

Co.,

is

more

than

85%

complete.

It

is

scheduled

to

be

delivered

to

the

purchaser,

Michigan Hospital

Service,

in

the autumn.

A

summary

of

its

main

features

follows (Note: "dd"

means

"decimal

digits"

and

"dds"

means

"decimal

digits

per

second"):

1.

Input

Convertor:

receives

source

data

in--':'--

punch

card

form

/

reads

twice,

compares,

edits,

arranges,

and

delivers

to

a

magnetic

file

unit /

processes

9QO

fully punched

cards

per

minute /

can

accept

the

presence

or

absence

of

anyone

of

12 punches

in

anyone

of

80

columns of

each

punch

card,

so

that

each

column

of

a punch

card

can

con-

vey

anyone

of 212

pieces

of

information

/

converts

into

binary

codes

in

fours

and

sixes,

giving a

mach-

ine

word

of 52

binary

digits,

which

can

be a

set

of

12 dd,

or

11 dd and sign,

or

8

alphanumeric

char-

acters,

or

anyone

of

many

other

choices

with

great

flexibility

.

2. Main

Memory

of Magnetic Tape: tape, 3

inches

wide / 31

parallel

recording

channels

/

reels,

2700

feet

long /

maximum

capacity

of

each

reel,

37.2

million

dd /

reading

and

recording

speed,

60, 000 dds / up to 100

magnetic

tape

file

units

can

be

directly

connected

in

a

single

system

/ up to 10

tapes

can

be

scanned

simultaneously

/

maximum

searching

rate,

600,000

dds.

3.

Intermediate

Memory

of Input

Register

Buf-

~:

magnetic

cores

/

total

capacity, 1488 dd /

receives

data

at

rate

of 60, 000 dds /

delivers

data

to

high-speed

memory

at

420, 000

dds.

4.

High-speed

Memory

of Magnetic

Cores

/

24,000

dd

capacity

/

access

in

parallel/access

- 8 -

time

for

a

machine

word

of 12 dd, 10

microseconds.

5.

Operations

in

the

Central

Processor:

addit-

ions,

4000

per

second

/

multiplications,

1000

per

second./

speed

of

sorting

machine

words

of 12 dd,

5000

per

second,

or

60, 000 dds /

speed

of

merging

machine

words,

5000

per

second,

or

60, 000 dds /

thus

for

business

and

commercial

problems

the

machine

can

utilize fully

the

speed

of

the

magnetic

tapes.

6.

Intermediate

Memory

of

Output

Register

Buffers:

magnetic

cores

/

total

cagacity,

1488 dd /

receive

data

at

420, 000 dds /

record

on

magnetic

tape

at

60,000

dds.

7. Output

Converter:

reads

data

from

magnetic

tape

at

60, 000 dds /

delivers

80

column

punch

cards

at

100

per

minute

/

prints

either

150

or

900

lines

per

minute, of 120

characters

each.

8. Checking,

etc.

Completely

checked

internally

by

a

built-in

checking

system

/

has

an

extremely

.

versatile

system

of

orders.

*---------------------------------------*

THE

PROJECT

STRETCH COMPUTER (IBM)

Nell

D.

Macdonald

New York,

N.

Y.

The

Los

Alam'os Scientific

Laboratory

of the

U.

S.

l\tomic

Energy

Commission

contracted

in

November

1956 with

International

Business

Mach-

ines

Corp.

for

a

computer

100

to

200

times

as

fast

and

as

capacious

as

the

most

powerful

commercial

computers

of the

present

day,

such

as

the

IBM

70.4

and 705. The

name

assigned

to

the new

project

was

STRETCH.

The

Atomic

Energy

Commission

has

begun

to

design

reactors

and

other

apparatus

so

complicated

that

one

hundred

billion

arithmetical

operations

may

New

Products

and

Ideas

be

required

to

make

one

evaluation

of

a

single

de-

sign.

A

present-day

computer

would

take

six

mon-

ths;

the

Project

Stretch

computer

is

expected

to

take

a

day.

Some

of

the

main

directions

for

pushing

toward

this

increase

of

speed

and

capacity

are:

(1)

the

use

of

faster

components

coming

out

of

laboratory

work

on

scientific

frontiers;

(2)

anticipation

now of

solv-

ing

certain

problems

within

a

few

more

years;

(3)

multiplexing

equipment

to

gain

concurrent

operation,

thus

avoiding

waiting

by

the

central

computing

unit;

and

(4)

more

powerful

instructions

within

the

com-

puter

to

reduce

the

number

of

steps

in

calculation.

Some

information

about

the

features

that

this

computer

will

have

is

now

available.

The

Project

Stretch

computer

will

have

a

num-

ber

of

main

sections.

One

is

an

input-output

sec-

tion

to

maintain

smooth

swift

communication

With

the

magnetic

tape

file

units.

A

second

section

is

an

intermediate

serial

computer

for

editing

the

flow

of

input

and

output

data.

A

third

section

is

a

cen-

tral

parallel

computer

to

operate

as

fast

as

possible

on

the

assembled

calculating

work

of

the

problem.

The

machine

word

will

be

12

to

15

decimal

digits

long.

The

intermediate

serial

computer

will

operate

in

either

binary

notation

or

decimal

notation,

and

in

either

fixed

point

or

floating

point

arithmetic.

It

will

perform

addition

in

2

or

3

microseconds,

and

multiplication

in

5

to

15

microseconds.

Logi-

cal

operations

that

are

more

comprehensive

than

are

most

of

those

in

present

day

computers

will

be

carried

out

in

2

microseconds.

The

speed

of

the

intermediate

serial

computer

will

be

5

or

10

times

the

speed

of

the

IBM 704

or

705.

The

central

parallel

computer

will

consume

a

million

or

more

instructions

a

second,

which

will

be

prepared

for

its

consumption

by

the

intermediate

serial

computer,

which

in

its

turn

will

use

state-

ments

in

mathematical

and

logical

notation.

In

the

central

parallel

computer,

fixed-point

addition

and

subtraction

will

take

place

in

two

tenths

of

a

micro-

second.

Floating-point

addition

and

subtraction

will

take

place

in

six

tenths

qf

a

microsecond.

Multiplication

will

take

place

in

1.

2

microseconds.

Transfer

over

buses

will

take

place

in

two

tenths

of

a

microsecond.

The

types

of

memory

and

input-output

devices

will

include

present-day

types

such

as:

ferrite

core

memory,

magnetic

tapes,

paper

tapes,

man-

ual

keyboards,

typewriters,

card

readers,

card

punches

t

and

printers.

New

types

will

also

be

in-

cluded:

a

magnetic-disk

randomly

addressable

memory

capable

of holding one

million

machine

words

and

communicating

at

the

rate

of

one

word

every

four

microseconds;

and

new

magnetic

tape

units

communicating

at

nearly

the

same

speed.

The

total

memory

may

per

haps

be

one

hundred

million

words.

The

intermediate

serial

computer

will

have

a

high

speed

parallel

memory

of8192

words

of

64

hits

each.

Its

full

cycle

of

operation

will

be

two

microseconds

and

it

will

read

out

a

word

in

8

tenths

of

a

microsecond.

The

central

parallel

computer

will

have

a

high

speed

parallel

memory

of

512

words;

its

full

cycle

of

operation

will

be

half

a

microsecond.

The

machine

will

have:

automatic

checking; a

means

for

localizing

and

signaling

errors;

and

in

several

places,

automatic

correction

of

errors.

Transistors

in

the

machine

will

perform

at

a 10

megacycle

repetition

rate;

these

are

coming

out

of

IBM

laboratories

right

no}\'o

*------------------~--------------------*

PROGRAMl\1ING WITH SOAP

(SYMBOLIC OPTIMUM ASSEMBLY PROGRAMMING)

Lockheed

Missiles

Van Nuys,

Calif.

Employees

don't

have

to

be

computer

experts

to

use

giant

"electronic

brains"

at

the

Van Nuys

plant

of

the

Lockheed

Missile

Systems

division.

The

company

announced

on

March

19

that

they

will

have

the

opportunity

to

use

its

IBM 650

computers

on

a

do-it-yourself

basis

to

solve

mathematical

problems

they

encounter

in

their

work.

To

qualify,

all

an

employee

has

to

do

is

take

a

12-hour

course

in

programming

on

company

time.

A knowledge of

simple

algebra

is

the

only

prerequisite

for

the

course.

E.

K.

Fisher,

head

of

the

plant's

mathematics

and

computer

services

department,

said

the

new

opportunity

is

aimed

to

help

employees

with

mathe-

matical

problems

too

complex

to

be

solved

e~sily

with

desk

computers

and

slide

rules.

It

will

also

ease

the

work

load

of

his

professional

programmers,

who

can

spend

more

of

their

time

on

highly

techni-

cal

problems.

Fisher

estimated

that

an

employee

can

easily

program

and

get

in

a

day

the

answer

to

a

problem

which

would

take

him

a

week

to

solve

by

usual

means.

Here's

the

way

he

will

do it:

First.

(cont'd

on page

37)

- 9 -

CAM

PROFILE

DESIGN

WITH

THE

UNIVAC

120

Louis

D.

Grey

Remington

Rand

-Univac

Division

Norwalk, Conn.

A good

portion

of

the

work

of

the

Univac

Engineering

Division

at

the-South

Norwalk Lab-

oratory

of

the

Sperry

Rand

Corporation

concerns

itself

with

the

design

of

cams

for

various

input

and

output

devices.

These

cams

are

now

produced

by a

well

integrated

process

which

consists

of

mathematical

analysis

of

the

cam

profile

require-

ments,

construction

and

evaluation

of

mathemati-

cal

functions

which

embody

these

requirements

with

the

aid

of

the

Univac

1.20,

and

the

produc-

tion

of

a

paper

tape

from

the

punched

card

out-

put

of

the

computer

for

the

purpose

of

guiding

an

automatic

cam

cutting

machine.

This

process

has

resulted

in

a

savings

of

days

over

the

tedious

process

of

mechanically

drawing

the

cam

radii

and having

the

cam

cut

on

a

human

con-

trolled

machine.

In

this

paper

we

describe

the

problems_encountered

in

the

design

of

a

cam

profile

and

illustrate

the

uses

of

the

Univac

120

in

solving

this

problem.

Accordingly,

we

begin

with

a

description

of

the

Univac 120.

The

Univac 120

is

a punched

card

computer

using

90

column Remington-Rand

cards

and

capable

of

handling

alphabetic

or

numeric

data.

It

con-

sists

of

an

electronic

computing

unit

and a

sensing

and punching

unit

operating

at

a

maximum

speed

of

150

cards

per

minute. Data

is

processed

according

to

a plugboard program

consisting

of

40

triple

address

steps

having

the

following

format:

1.

First

value

(VI)

......

any

input-constant

or

storage

value

2.

Process

(V2)

....

addition,

subtraction,

multiplication,

division

3.

Second

val

ue

(V2)

•.•..

any

input-constant

or

storage

value

4.

Result

(R)

•....

any

intermediate-output

storage

except

one used

for

Vi

or

V2

5.

Plus

result

branching

6.

Minus

result

branching

Additional

steps

are

obtainable

through

the

use

of

any"

or

all

of

the

96

selectors.

Special

instructions

are

available

for

releasing'values

in

storage

for

punching

into

the

card,

sorting

cards

into

two

groups,

clearing

storages,

trans-

ferring

selectors,

etc.

The

basic

word

length

is

10 decimal

digits

with

a

22

decimal

digit

accumulator.

Gate

vacuum

tube

trigger

circuits

are

used

with

bi-quinary

coding.

Cold

cathode

gas

tubes

are

used

for

storage

with

a

slightly

different

type

of

bi-quinary

coding.

Process

time

including

an

automatic

inverse-process

zero

check

averages

10

milliseconds

for

addition

and

subtraction

and 50

milliseconds

for

multi-

-10

-

plication

and

division.

At

maximum

speed 325

milliseconds

of

computing time

are

available.

If

the

computation

takes

longer,

the

sensing-

punching

unit

automatically

waits

until

the

com-

putation

is

completed. There

are

three

types

of

storage,

as

follows:

A.

Intermediate-output

storage;

12

units

of

10

decimal

digits

each,

including

sign;

or

a

total

of

120 decimal

digits.

B.

108 (even)

or

216 (odd) decimal

digits

of

plugboard

constants,

which

may

be grouped

into

values

of

10

digits

maximum.

C.

120

digits

(maximum

of

90

or

anyone

card)

of

card

input,

which

may

be grouped

into

values

of

10

digits

maximum.

Branching

is

fully

flexible

and

takes

place

on

a

plus

or

minus

result.

U5E

OF

'!HE

UNIVAC

120

1N

CAM

DESIGN

A

cam

is

a

device

for

imparting

motion through

another

device

called

a

follower

which

is

always

in

contact

with

the

cam.

Motion

is

achieved

by

varying

the

lengths

of

the

radii

measured from

the

center

of

rotation

of

the

cam.

The

follower

cen-

ter

may

be

constrained

to

move

along

the

radius

of

a

cam

in

which

case

the

motion

is

said

to

be

radial

or

along

the

arc

of

a

circle

in

which

case

we

speak

of

non-radial

motion. These

moti~ns

are

illustrated

in

Figures

1 &

2.

The

cam

profile.

consists

of

a

series

of

smooth

curves

which

incorporate

the

dynamic

char-

acteristics

of

the

system and which have

identical

displacements,

velocities

and

accelerations

at

the

point

where any

two

curves

are

joined.

Well

known

curves

used

in

the

design

of

cam

profiles

qre

the

harmonic and

cycloidal

curves(l).

In

general,

however, a

profile

can always be

obtained

in

the

form

of

a

series

of

polynomials each

of

whose

degree

is

one

less

than

the

number

of

conditions

the

polynomial must

incorporate.

Thus suppose

that

a polynomial

in

9

is

to

have

given

displace-

ment,

velocity,

acceleration

and

rate

of

change

of

acceleration

characteristics

at

Q

~

0 and

at

9 =

~/.

We

shall

suppose

the

follower

center

to

have

radial

motion

since

the

results

need very

little

modification

for

non-radial

motion.

Let

t'Cp=

Ao

(~J

+

A,{:J'

-I-

A"

(:,r

-I-

A3

(~.r+

Aq

(~.Y

-I-

As

(;S

+

A,(

:,)

+

A7

Cam

Profile

Design

with

{Co)

:

C.

{"(o)

-=

<=s

{

Cr,)

:::.

Ca...

{"

(PI)

=-

C,

{'

(c) = C3

{hI

(0) =

C7

{'

(t,)

=c:::"

~~~

(r,)

:.

Cs

Imposing

the

above

conditions

leads

to

a

system

of

eight

simultaneous

linear

equations

which

are

then

solved

to

determine

the

A's.

These

equations

are

conveniently

solved

on

the

Univac

120

by

employing a

routine(3)

based

on

the

method

of

direct

elimination

featuring

floating

point

decimal

arithmetic

and good

for

matrices

up

to

(15x15).

If

desired

the

system

of

equations

can be

solved

for

various

sets

of

constants.

For an ill x ill

matrix

(m~15)

the

total

number

of

cards

involved

is

~

m

(m

+

1)

On

+

2n}

where n

is

the

number

of

columns

of

constants

that

the

equations

are

to

be

solved

for.

The

computer time

necessary

to

solve

these

m

sets

of

equations

is

J2

m

(m

+ n

(m

t 2n)

Average

card

speed

The

average

card

speed

is

dependent

upon

the

elements

of

the

coefficient

matrix.

Having

solved

these

equations,

what

is

then

desired

is

the

eval

uation

,of

the

resul

ting

polynomial

at

small

intervals,

say

~o.

This

gives

the

radii

of

the

cam

measured about an

axis

of

rotation

through

the

center

of

the

cam.

The

polynomial

is

conveniently

evaluated

on

the

Univac

120

by

using

Horner's

algorithm.

Thus

for

{(G)

=

Ao

(~.)""'

+

A.(~,)-'+

....

+

A_

J'

=

~

-~o

p~

{ea+

' = e

{~

+

A~+,

, t

'C

0,')

••.• 1m-I

WhicnIequires

~

additions

and m

multiplications

to

get

f(G).

Permanent

charts

of

radii

for

vari-

ous

angles

can be

made

up

if

a

curve

is

used'fre-

quently.

The

results

are

produced

in

the

form

of

punched

cards.

Information

is

then

taken

off

the

card

and

either

tabulated

on

special

drawing

paper

enabling

a

draftsman

to

construct

a drawing

of

the

cam

profile

or

converted

in

the

form

of

a

paper

tape

by

means

of

the

Remington

Rand

Card-to-Tape

Converter

(Model

318).

The

resulting

paper

tape

output

is

then

fed

into

an

automatic

machine

tool

which

cuts

a

cam

having

the

specified

profile.

It

has been found

that

cams

of

high

quality

can

be produced

by

this

process

cheaper

and

faster

than

had

previously

been

possible.

BffiLIOGRAPHY-

1 -

"Plate

Cam

Design",

M.

Kloomok

and

R.

V.

Muffley,

"Product

Engineering",

February

1955, p.

156

2.

,...

"Plate

Cam

Design",

M.

Kloomok

and

R.

V.

-11

-

Muffley

"Product

Engineering",

May

1955, p.

155

3 -

"Simultaneous

Linear

Equations

and

Matrix

Inversion

(floating

decimal method)

with

Univac

120

punched

card

computer~

Remington Rand, 315

Fourth

Avenue,

New

York,

N.

Y.

4 - "Using Punched Cards

for

Automatic

Machine'~

A.

H.

Stillman,

"Product

Engineering",

June

1955, p.

172

5 - "Polydyne

Cam

Design",

D.

A.

Stoddart,

Machine Design, Vol.

25)

January,

February,

March

1953, pgs. 121-135, 146-154, 149-164.

FOLLOWER

0..

Figure

2

SUPPLEMENTING

ELECTRONIC

EQUIPMENT

WITH

A

MODERN

COMMUNICATIONS

SYSTEM

Monroe

M.

Koontz

Inland

Steel

Co.

East

Chicago,

Ind.

ffieprin'ted

by

permission from

"N.

A.

C.

A.

Bulletin",

vol.

38, no.

5,

Jan.

1957,

published

by

the

National

~ssociation

of

Cost Accountants,

New

York,

N.Y.>

The

paperwork

structure

upon which

con-

trol

is

based today does

not

have the

physi-

cal

attributes

needed

in

the

transition

to

an

eight

hundred

billion

dollar

economy

in

1957.

Growth

of

such magnitude

means

continued

de-

centralization

and the accompanying

increase

in

producing

units.

Each of

these

must be

able

to

communicate

to·produce

efficiently.

Overhead

costs

will

increase

geometrical.ly

unless

the

basis-paperwork-of

our

present

control

technique

is

eliminated

in

large:chunks.

It

is

intended,

in

this

article,

to

relate

the

new

array

of

office

machines

with

the

elements

of

control,

to

outline

a broad

perspective

for

improving

their

application,

and

to

suggest

changes

in

emphasis, with

regard

to

office

automation, which

may

be

desirable

to

avoid

what might be

called

a

plateau

of

development,

i.

e.,

the

limitation

of

development

to

a fixed

level.

In approaching

automati~n

of

the

office,

many

experts,

busily

applying

well

advanced

knowledge

in

organization

.management and

trad-

itional

communication

techniques,

seem

to

have

over looked

the

startling

.poss

ibi

Ii

ty

that

(ilys-

ical

communication

facilities

now

in

general

use

have

become

so

old,

technologically

speak-

ing,

that

they must be

on

the

verge

of

super-

annuation.

Paradoxically

-because the

line

of

progress

for

management,

itself,

lies

in

de-

centralization--the

best

commercial

potential

of

information

processing

system

may

lie

in

centralization

or

semi-centralization

of

com-

munications and

data

processing.

Decentrali-

zation

of

industry

puts a

strain

on

over-all

control

and

also

places

a premium

upon

control

at

production

locations.

This

dual

need can

be

served.

The

executive

span

of

control

can

be extended

considerably

by

new

electronic

communications, and the massive

ability

of

cen-

tralized

electronic

systems

to

boil

down

rou-

tines

full

of

detail

can be put

to

use

at

all

levels.

-12

-

Limited

Means

of

Control

Limits

the

EconomY's

Growth

Potential

Our

thinking

should

not

be burdened

by

undue emphasis

upon

tradition

either

in

de-

vices

or

organization.

Current

and

forth-

coming developments

will

be

best

adapted

to

use

in

terms

of

their

inherent

characteris-

tics,

and

these

will

impose a need

for

organ-

izational

as

well

as

technical

changes.

The concept most

vital

to

future

planning

is

the

analogy between sensory

functions

of

the

individual

and what might be

possible

in

the

office.

In

the

light

of

this

concept,

exist-

ing

organization

concepts

will

metamorphose

extensively.

(After

the

change,

the

game

of

organizational

chess

may

be resumed.) There

is

urgency.

here.

If

communications

are

not

given a

new

structure

with a

specific

goal

of

tremendous

acceleration,

economical use

of

high-speed

computers

will

be hampered and

predicted

business

growth

may

achieve

less

than

optimum

results,

and might outrun our

ability

to

control

it

with paper,

with

par-

tial

stagnation

following.

The

central

problem

is

that

the

span

of

business

control

is

circumscribed

substantially

by

paperwork. Accountants and managers

know

this

well

and admit

it

reluctantly.

To

reduce

the

burden

of

necessary

detail

by

electronic

communication and

processing

will

augment

the

talents

of

our

limited

supply

of

managers and

analysts,

thus

elevating

the

ceiling

over

new

business

growth.

Control

as

it

is

known

today grew out

of

the

industrial

revolution.

As

technical

prob-

lems

of

produc

tion

were surmounted

in

the

first

half

of

the

nineteenth

century,

the

commercial

consequences were

virile

expansion,

an

exten-

sion

of

division

of

labor,

and

continued

evo-

lution

into

departmental

allotment

of

func-

tions.

One

correct

term

for

this

process

is

Supplementing

organization,

in

itself

a

control

tool.

Organ-

ization

in

turn

encouraged

prolific

growth, be-

cause

it

lifted

the

limitations

of

single

man-

agement from

the

application

of

new

technology

potentials.

By

1900,

more

than

organization

itself

was

needed.

The

size

and expanded

log-

istics

of

business

were

getting

beyond

the

personal

grasp

of

the

management

representa-

tive.

The

medium

to

supplement-and

in

turn

to

constitute--his

control

was

paperwork:

reports,

charts,

and a body

of

clerks

to

produce them.

The

industrial

accountant

emerged

simul-

taneously

to

help

decision

makers

screen

vital

facts.

Paperwork

tools,~hile

not

exclusive

to

the

industrial

accountant,

do

largely

charac-

terize

his

activity,

because they

include

or-

ganization

charts,

the

card

of

accounts,

and

the

structure

of

reports

to

management.

The

accountant

is

traditionally

oriented

to

paper-

work and,

in

some

situations,

new

paperwork

is

the

only

way

to

get

control.

However,

it

has

natural

limitations.

These

are

time

factors

(in

terms

of

both

cost

and delay)

for

prepa-

ration

and

transmission,

plus a

relatively

moderate space

for

filing.

Until

the

last

few

years,

the

volume

and

complexity

of

busi-

ness had

not

progressed

sufficiently

for

these

limitations

to

hurt

the

operation,

but

the

in-

exorabl~

establishment

of

mUltiple

new

communi-

cation

lines-paper

vehicles

for

information-

each time a

new

center

of

activity

was

set

up,

finally

began

to

take

its

toll.

The

Essence

of

Control

Is

Immediate Reaction

It

is

a

real

dilemma. Communication

fa-

cility

must be improved

for

continued

busine~

growth

but,

in

the

natural

limitations

of

paperwork,

will

not

change.

The

best

we

can

hope

for,

without

a

basic

divorcement

of

com-

munications from paper,

is

to

palliate

the

effects

by

simplification.

However, the

Amer-

ican

economy

of

1975 can be commensurately

profitable

only

if

the

controllers

of

busin~

go

back

to

true

fundamentals

to

achieve

the

controls

they

are

seeking.

Basically,

these

fundamentals, long pushed

out

of

sight,

re-

semble

in

nature

the

original

control

endeav-

ors

of

an

owner

of

a

business.

On

the

current

day

scene,

they

take

form as

electronosensors,

or

the

extension

by

mechanical and

electronic

facilities

of

human

sight,

hearing,

and

mem-

ory,

all

interacting

instantaneously

to

rele-

gate

details

and

synthesize

action.

All

of

the

necessary

facilities

with which

to

enact

this

concept

of

control

are

on

the

market

to-

daY,and a combination

of

these

with

superior

potentials

can be

'cheaper,

more

reliable,

and

more

effective

within

a given

business

unit

than

anything

based

upon

messengers, pneumatic

-13

-

tubes,

or

medium

distance

wire

transmission,

such as telephone and

telegraph

extensions.

These

substitute

facilities

will

be

considered

in

greater

detail

and

the

use

of

true

electrono-

sensors

will

be

explained.

First,

however,

we

need

to

restore

to

our consciousness

the

orig-

inal

model

of

control

appraisal

and

action

as

it

existed

in

the

sole

owner

of

the

one-time

typical

small

enterprise.

The

successful,

if

archaic

proprietor,

~

a

data

originating,

processing

and

control system

within

himself.

His shop

contains

most

of

the

elements

of

the

business-

fixtures

and

inven-

tory,

a drawer

or

two

containing

his

purchase

and

sales

slips,

and the hidden strongbox

or

stocking

containing

the

rest

of

his

net

worth.

His

brain

serves

at

once

as:

1.

The

reporting

system which corresponds

to

input

in

a

modern

machine

installa-

tion.

2.

The

data

processing

center

containing

sorting

and merging

abilities,

pro-

grams

for

economical

purchasing,

pay-

ment

of

bills,

and

preparation

of

tax

returns.

3.

The

reporting

system which corresponds

to

output,

as

it

is

popularly

termed

today.

4.

The

evaluation

system, which

recogn~es

situations

and

generates

action

to

stop

pilferage,

control

waste,

and develop

policy

based

upon

sales,

cost

and oper-

ating

experience.

From

this

oversimplification,

several

secondary

considerations

will

place

in

per-

spective

the unsound claims

that

a

multiply-

ing

vqriety

of

parallel,

inadequate,

and,

for

the

most

part,

excessively

costly

developments

in

current

management

reporting

and

control

technology

will

solve

our problem. These

secondary

considerations

emphasize the

instan-

taneous

character

of

the

original

sole-owner

control.

We

might

visualize

him

talking

to

another

or,

by

extension,

we

might imagine any

number

of

proprietors

or

data

originating

and

processing

centers

"talking"

to

each

other.

In

the

sole

proprietor's

bio-mechanical system,

1.

There

is

visual

contact

with

input.

2.

There

is

auditory

or

hearing

~ontact

wi

th

input.

3.

Communication

is

instantaneous.

4.

Processing

capacity

depends upon the

size

and speed

of

the

memory.

5.

Reports

are

drawn

directly

from

mem-

ory

for

the most

part,

rather

than

being

transcribed

to

paper

for

tedious

and delayed

perusal.

6.

Action

may

,be

taken

quickly

without

paperwork.

7.

Input

is

largely

accurate

in

the

first

three

above,

by

reason

of

their

inher-

ent

value

to

each

other,

but

error

Computers

and

Automation

correction

is

poor

in

the

rest.

8.

The

element

of

top

managerial

surveil-

lance

may

intervene

at

any time

in

every component

of

the

system.

Expensive (and Doubtful) Developments

in

Data Transmission

The

quest

for

ways

in

which

to

utilize

advanced equipment

for

better

control

communi-

cations

goes

on

actively

but

not

always

effec-

tively

or

along

economical

lines.

Hundreds of

engineers

and management

consultants

across

the

country

have been

sinking

millions

of

dollars

into

adapting

known

telephonic

and

telegraphic

methods, along

with

paper

tape

or

punched

card

originating

equipment,

to

computer

input.

How-

ever,

most

of

these

devices

lack

vision,

hearing,

and

instantaneousness

in

varying

combinations or

altogether.

Furthermore, they

really

increase

paperwork because they

require

paper

to

bridge

from one

to

another,

and

this,

in

turn,

im-

poses need

for

more

hardware.

The

hardware,

itself,

has

generated

further

input

errors

from

compound

circuitry

and

relays

which sometimes

fail.

When

errors

occur,

paper

tapes

impose

exceptional

difficulty

in

locating

the

point

in

question.

A

large

insurance

company

indi-

cates

that

it

has

discontinued

paper

tape

transmission,

for

this

reason,

to

return

to

cards.

One

supplier

has announced a machine,

or

rather

a number

of

machines

circuited

to-

gether,

to

produce

cards

and

reports

at

vari-

ous

production

points,

but

this

system

lacks

vision

and

instantaneousness

,and,

to

couple

it

with

logical

(computing)

ability,

messengers

will

have

to

be

hired

to

pick

up

the

cards

at

each

unit.

Such a

device,

which

cannot

poss-

ibly

be

used

full-time

on

a

decentralized

basis,

rents

for

more

per

month

than

the

wages

of

an

average

clerk.

There

are

other

still

more

expensive

ways

to

communicate.

Even

these,

because

communi-,

cation

to

a

data

evaluation

center

is

so

vital

to

control,

have been

pressed

into

service.

One

such

means

is

represented

by

pneumatic

tubes,

which

cost

up

to

several

hundred

dollars

per

installed

lineal

foot.

They

are

all

but

inflexible,

once

set,

and

fixed

charges

for

a

substantial

installation

will

equal

the

wages

of

a number

of

people.

They

also

lack

sound,

which must

be

supplied

by

telephone

auxilaries,

and they

lack

the

visual

component

altogether.

Being

built

to

transport

physically

full-scale

paperwork,

with

its

clerical

preparation

and

timelag

straits,

they have

relatively

low

transmission

capacities.

Computers, Magnetic

Storage

-and

the

In

pu

t Problem

Moreover, communications must tend

to

-14

-

some

point,

or

points,

of

collection

and

eval-

uation.

Hence,

at

this

juncture,

the

data

pro-

cessing

facility

should

be

considered,

i.

e.,

how

to

use communicated

data

for

comprehension

and

reaction.

On

a

warship,

this

facility

is

a

group

of

men

and

technological

nerves

to

all

quarters,

termed a combat

information

center.

In a

plant,

it

is

designated,

variably,

after

many

decades

of

progressive

decentralization,

as

an

office,

whether

mill,

accounting,

pro-

duction

control,

shipping,

or

some

other

de-

scriptive

adjective.

Superintendence

of

these

offices

has been

centralized

in

waves as

the

spreading

out

process has caused them

to

be-

come

identifiably

functionalized

and,

in

this

process,

establishment

or

expansion

of

a

staff

function

has

coincided

with

each

new

wave. In

all

cases

there

is

an

input,

mostly

of

paper-

work. Communication

channels

are

plant

mail,

telephone,

pneumatic

tubes,

and

conferences.

In a few

recent

cases,

paper

tape

is

the

medium.

And,

like

the

old

telephone

company

advertise-

ment,

es~ablishment

of

one

new

instrument

where

ten

existed

before

required

ten

new

lines

of

liaison,

except

that

the

lines

have been

lines

of

paper

with

clerks

at

both

ends.

Electronic

computers have

entered

this

scene and

are

operating

successfully

in

spite

of

the

paperwork load

at

the

input

end

but,

so

far,

the

most

effective

applications

have been

integrations

of

master

data,

which avoided a

new

flood

of

reports

coming

in.

The

latter

are

better

absorbed

into

an

integrated

system

over the time needed

to

redesign

the

basis

of

origination.

Detailed

liaison

is

needed

all

the

way

to

the

source

in

this

endeavor.

Once

the

paperwork has been adapted

to

the

purpose,

it

can

be

readily

filtered

by

a

junior

clerk

with

a keypunch machine

so

that

the

computer

may

be

set

up

to

handle

the

data.

Data punch-

ing

on

this

,basis

is

duplication

of

effort,

again

using

paper,

but

the

costs

are

more

than

overcome

by

some

integration

in

processing.

Thus,

the

computer

is

no

problem as

far

as

its

internal

operation

is

concerned

but,

in

draw-

ing

more

paper through

the

door

of

the

machine

room,

the

communication problem

is

accentuated.

And

the

key-punch group becomes packed

tighter

than

office

standards

permit.

In

the

light

of

these

factors

the

present

and

potential

economics

of

computers must

be

considered.

What

processing

problems do they

solve?

Like

the

sole

proprietor,

they

com-

municate

rapidly

internally.

They have

sub-

stantial

processing

capacity.

Reports

may

be

drawn

directly

from

memory

with

moderate and

ever-improving

speed

for

management

action.

The

problem

of

error

correction

now

lies

most-

ly

outside

the

system, once

it

is

"debugged".

Input

both

external

and

internal

to

the system

and

their

very

interrelationship

develop

as

joint

and

several

integration

factors

affect-

ing

computer

capabilities.

The

integration

Supplementing

of

production

reports

from

two

adjacent

units,

for

example,

means

that

the

computer can

auto-

matically

produce

inventory

data

between

units

and,

usually,

more

than

half

of

the

external

repetitiveness

in

data

origination

can

be

elim-

inated.

If

production

data

be

stored

in

a

ran-

dom

access

magnetic

file,

even

the

inventory

reports

can

be

eliminated,

for

all

inventory

data

is

available

instantaneously

and

visually

to

anyone

who

has

an

interrogating

connection

between

his

office

and

the

magnetic

storage

unit.

If

production

orders

were mechanized

upon

receipt,

the

external

input

concerning

each

order

from each

production

process could

be

reduced about

65

per

cent.

l

Random

access

on

the

foregoing

terms

is

a

key

point

here.

Announcements,

since

this

paper began

in

preparation,

have been most

sig-

nificant

in

terms of opening up

the

large

ran-

dom

access

memory

capacity

to

provide

memory

characteristics

like

those

of

the

sole

propri-

etor.

It

appears

that

the

economy

of

magnetic

storage

is

already

getting

to

a

point

at

which

it

will

appear cheaper than

cards

on

many

jobs,

but

caution

in

applying

it

is

still

needed.

Still

other

recently

announced developments

are

similarly

indicative.

One

laboratory

has

a two-ended cathode

ray

tube

with

a

metallic

grid

for

permanent

storage,

capable

of

hold-

ing

one

million

bits

on

a

square-inch-screen.

Its

speed

is

great

because one end

reads

while

the

other

end

writes,and

the

screen

is

so

small.

Such a development in

production

would

render

obsolete

discs

and

drums

as

to

speed,

space

requirements,

and

capacity,

and even

this

would

ultimately

be

excelled.

Nuclear

physicists

chat

about

the

ideal

as

calculation

and perhaps

stor-

age

in

a

matrix

of

differentially

chargeable

molecules.

Not only would computers

built

on

such

principles

exceed

by

a

factor

of

thous-

ands

anything

available

today; they would

be

ridiculously

compact

for

their

capacity,

and

possibly

inexpensive

after

perfection

of

the

necessary

manufacturing

processes.

The

upshot

is

that

commercial memories

will,

in

the

near

future,

be

capable

of

relegating

the

detailed

input

of

the

business

to

storage

for

reference

by

the

computing component as needed,

just

as

in

the

sole

proprietor's

mind. But the

tool

will

be expanded

to

the

proportions

of

the

economy

as

we

know

it,

and

it

will

be

more

capacious,

more

speedy and

potentially

less

erroneous

than the

proprietor

ever

was. That

is

fine.

However,

its

very

instantaneousness

is

also

silly

if

messengers

are

coming

in

eight

to

twenty-four

hours

late

with

the

input.

IBased

on

a

survey

of

two

production

centers

made

for

purposes of

this

discussion.

The

percentage

will

vary depending upon the

sys-

tem

and

the

degree

of

integration.

There

is

little

purpose

in

not

glvlng

large

scale

instant

storage

an

input

when

the

event

occurs,

and

this

is

the

strongest

argument

against

large

centralized

computer

facilities

until

communication

facilities

can

be

improved.

Another

factor

of

uselessness,

i.e.,

ir-

relevance

of

advantages

of

speed and

capacity,

inheres

in

any

lack

of accuracy

in

input

as

long as computers

are

expensive.

With

an

I.

B.

M.

650,

for

example,

input

may

range

up

to

a

maximum

of

over 900,000

digits

per

hour.

Even

if

only'three

of

these

are

wrong'enough

to

stop

the

machine and

it

takes

only

two

min-

utes

to

correct

each one,

10

per

cent

of

the

capacity

has been

lost.

On

larger

machines

this

item

is

so

important

that

some

installa-

tions

have a machine

scheduler

(the

705 does

over 40,000

logical

operations

per second)

to

enforce

nondelaying

procedures.

If

an

error

stops

the machine

more

than a very

few

minutes,

the

memory

must

be

dumped

by

the

hapless

oper-

ator

so

that

the

in

line

may

get

to

his

problem.

The

price

of

delay

is

over

$150 per hour.

These

considerations

reinforce

the need

for

development

of

reliable,

inexpensive,

plant-wide

data

recording

and

transmission

systems,

which can provide audio and

visual

contact

with

low

origination

error

ratio.

It

appears

that

such systems

also

may

be

the

key

to

decentralization

of

memory

elements

of

a

unifiable

computer complex

in

various

loca-

tions.

For example,

if

a number

of

memory

devices

can

be

hooked

electronically

into

central,

their

distance

from

the

unit

is

mere-

ly

the

length

of the

connection,

and

distance

does

not

occasion the

delays

of

traditional

input-output

via

paper

linkages.

It

is

the

instantaneous

merging of

plant-wide

data

that

matters.

Nevertheless,

the

best

commercial use

of

a computer

installation

currently

implies

centralization

because:

1.

Personnel

acquainted

with

the

interaction

of

computer and commercial systems

are

too

scarce

and too expens i

ve

to

Its

pread

thin"

•

2.

The

new

category

of

communications-data

originating

control

expert

which

will

have

to

be

brought

into

existence

in

the

manufacturing

establishment

does

not

ex-

ist,

although

materials

are

available

to

start

training.

3.

A

large

memory

is

not

yet

adequately

adaptable

to

decentralization,

which

would remove

it

from the computing

unit

and

the

master

file

systems.

A

large

magnetic

memory

is

needed

for

paperwork

elimination,

rather

than

simplification,

and

elimination

occurs

when

master

files

are

integrated.

Presently

applied

com-

munications

are

not

adequate,

either

in

-15

-

Computers

and

Automation

terms of

facilities

or personnel,

to

as-

sure

high-speed

integration

of

decentral-

ized

memory

components with the

presently

developed

heart

of the system,

i.e.,

cards,

tapes,

and

programs.

4.

Integration

of

data

simul-

taneously reduces

input

error

ratios

and

reduces the need

for

duplicating

data

from

external

sources.

5.

The

reduction

of

external

input

gained

by

centralized

integration

would reduce

the load

on

the communication system,

rendering

it

more

effective

for

a given

outlay.

6.

Concentration of

data

improves computer

payoff

and

builds

necessary knowledge

of

techniques in the

organization.

The

loading of a medium-size machine

must

be

effected

rapidly

to

justify

the econ-

omy

of

larger

equipment. Cost per

cal-

culation

on

a loaded

705

would

be

less

than

10

per

cent

of

that

on

a loaded

650,

in

terms of

rental

alone.

7.

Larger

capacity

units

handle

greatly

more

complex

logistical

and

research

problems, which provide

greater

payoffs

than

straight

commercial

applications,

to

supplant

today's

generally

unsophis-

ticated

analyses.

8.

Complaints

already

have been heard

that

all

of the

fastest

computers

available

cannot handle computer demand,

either

in terms of

volume

{this

is

on

a

nation-

al

basis} or in terms

of

problem

com-

plexity

•.

ftdvancement

of

the

art

is

cer-

tain

to

bring

more

demands.

A Proposal

for

Control Communications

by

Monitored

Television

The

limitations

in

data

transmission

and

input,

along with

other

limitations

which have

just

been

described,

would

be

discouraging,

were

it

not

that

modern

components

are

avail-

able

to'develop

a

reporting

system lower in

cost

than any complex of

traditional

paperwork

adapted

to

50-year-old

communications

technol-

ogy

to

ease

its

burden. Such a system can

command

the

economy

of

integration

in

a

high-

speed processing

organization,

provide channels

for

extremely heavy volumes

of

transmission,

and

provide

audio-visual

contacts

among

many

points.

The

proposal involves

television

cam-

eras

for

data

transmission.

They

may

be

ob-

tained

for

prices

under $1,000 per

unit,

ranging

up

to

two

or

three

times

that

amount

for

units

having combinations of remotely con-

trolled

multiple

turret

lenses,

sound,

and

weather

protection

components.

Scanning tube developments

in

recent

months

already

have reduced

prices.

The

in-

dustry,

by

means

of

printed

circuits,

other

-16

-

innovations,

and

volume

demand

for

industrial

applications,

undou'btedly could

cut

prices

50

per

cent

in

coming

years.

Units

may

be

port-

able.

They

require

no

programming as

in

the

case

of tape or card punching gadgetry_ Co-

axial

connections from cameras

to

monitors

at

about a

dollar

a

foot

would

pe'

far

shorter

and

cheaper than

multiple-circuited

mechanical

data

originators

wired

to

messenger

service

or

connected

all

the

way

to

central.

Reli-

ability

of camera operation

is

superior

and

maintenance simple.

If

an

unusual mainten-

ance problem occurs, another

unit

may

be

plugged

in

while

repairs

are

made.

In

contrast

to

tape-punching equipment,

which

costs

$2,000

to

$4,000 per

originator

and

receiver,

a

variety

of

forms can

be

trans-

mitted

by

camera without programming, thus

rendering

floor

da'ta

simi

taneous

ly

adaptable

to

local

use

and

data

transmission

within

reasonable

limits.

The

need

for

originators

would

be

greatly

condensed.

If

a

number

of

stations

are

multiplexed

to

the monitor,

ml-

tiplexing

is

electronic

and

visually

instan-

taneous, whereas

older

devices

multiplex

me-

chanically

and

far

more

slowly.

These charac-

teristics

are

vital

to

high-speed, high

volume

data

origination

and

processing,

and

also

re-

store

the opportunity

for

visual

management

surveillance

of

operations.

A

description

of

what should happen

to

data

will

make

the proposal

concrete.

Most

readers

will

have observed

that

data

origina-

tion

is

only a

part-time

job

for

many

produc-

tion

recording

clerks,

although they

are

writ-

ing

their

reports

in slow longhand. Because

of

this

fact,

a

semi-centralized

monitoring

station

{which

would

also

really

be

a decen-

tralized

outpost

of machine accounting}, should

be

able

to

handle in

practical

fashion

the data

from ten or

more

points

in

the

plant.

For

ex-

ample, a

common

work

situation

at

present

is

a

recording

60

per

cent

to

waiting

40

per

cent

ratio.

Based

upon

the survey

referred

to

in

a

previous paragraph,

65

per

cent

of the

data

formerly

~andwritten

at

all

points

would

be

in

computer

room

master

files

-

in

the

integrated

system-

and

this

would

cut

the

data

originating

factor

required

for

each

point

to

20

per

cent

at

the

monitor.

At

the monitor, the

usual

three-for-one

speed advantage

of

keyed over

handwritten

data

would apply, thus reducing

the

data

origination

time

for

each

point

moni-

tored

to

an

average

of

6

2/3

per

cent.

This

estimate

is

conservative,

because

it

has

not

been

factored

further

to

adjust

for

better

skilled

personnel,or

the enhancement

of

speed

which occurs

in

the contemplated range

of

punches per item as a consequence of fewer

keys

activated

per

transaction,

or

for

addi-

tion

of

totals

formerly done

by

the

mill

re-

corder,

as

part

of the

original

60

per

cent

Supplementing

work time,

and

now

done in the computer.

It

is

seen

that

data

relaying

and

visual

switching

from the monitors

will

depend

upon

a type

of

control

communications personnel

generally

new

to

industry.

To

obtain

the

necessary accuracy

for

unimpeded

volume

of

external

input

to

the computer,

it

also

is

probable

that

these persons would perform

every

entry

twice

to

obtain

verification

on

the

spot

before

triggering

their

transmissions.

However, except

for

salary

levels,

these

per-

sons would

cost

nothing

if

their

work were

loaded

properly.

They

would

offset

equiva-

lent

personnel otherwise employed

in

the

ma-

chine room.

Training

of monitor personnel

would correspond

to

that

of,

say,

railroad

telegraphers

or

nautical

radiomen,

and

would

cover

data

origination

and

control

techniques,

operating

practices

at

points

scanned

in

the

local

complex, and,

to

some

extent,

the

com-

puter

programs

for

processing

its

particular

data.

The

personnel would

service,

in

re-

verse,

any

data

from the

central

unit

requested

for

local

analysis

or

control

purposes.

It

is

likely

that

the

data

transmission

component of

carrier

waves

from monitors would

preferably

use

key

punching equipment

or

an

adaptation.

So

far,

it

is

better

than

other

types of

de-

vices

in

reliability

of

operation.

Monitors

further

would need

policy

enforced

control

or

disciplinary

procedures 9ver the

stations

under

their

observation.

Completed

entry

and

verification

of

weights,

counts,

and

simple

order

,indicia

would

be

signalled

from monitor

to

camera

by

a green

light,

advising

the

material

handler

that

production has been

recorded.

From

the

monitor

to

the computer, a

private

micro-wave

transmission

system (these

units

also

contain

the

camera

multiplexing

circuits)

can

be

en-

gineered economically

for

all

but

small

plants.

Simultaneous

visual

and

data

transmission

on

one

carrier

beam

is

obtainable

either

by

"band

splitting,"

which

is

using

parts

of the

trans-

mitter's

wave

band

for

different

channels, or

by

use of

extra

transmitters.

Thus, manage-

ment

could

get

visual

contact,

with

one

or

more

points

per

transmitter,

without

inter-

rupting

flow of

data

to

the

processing

machine.

A

transmitting-receiving

combination with

the

necessary

multiplexing

system

reportedly

is

available

for

$30,000 or

less,

although

elaborate

units

cost

more.

If

ultra-high

frequencies

(up

to

10,000 megacycles

are

available)

are

used,

signals

may

be

beamed

thirty

miles with

ridiculously

low

power-

less

than

10

watts

-and

by

using

only moder-

ately

high towers.

The

plant

roof

is

often

high enough

to

mount

the antenna without con-

structing

a tower.

The

high

frequencies

lie

outside

the range

of

static

for

all

practical

-17

-

purposes.

Compared

with

this

medium,

pneumatic

tubes

are

poor. With the

latter,

coverage

of

a

few

points over a

maximum

range of

1000

feet

may

cost

over $100,000

and

have lower

trans-

mission

capacity.

The

wavelength of a 10.000

megacycle frequency

is

about

l~

inches.

Trans-

mitting-receiving

antennas

are

"dishes"

as small

as a

few

inches

in

diameter

and

precisely

aimed

at

each

other.

Incoming

data

at

the processing

location

would

be

recorded

either

on

film

(tapes) or

go

directly

to

random access

storage.

The

data,

although

verified,

might

be

occasionally

erron-

eous, but a small

residue

of

error

can

be

cor-

rected

to

some

extent

by

the processing

machine

because of the

integration

of the system.

Stock checks

on

the production

floor

would

be

minimized, compared with the present-day

situ-

ation.

Most

day-to-day

reports

to

management,

reduced

by

the programming of

the

system

to

exceptions,

would

be

drawn

from the

process-

ing

machine

in

summary,

and

visual

form only.

Details

would

be

fed out

to

operations

visu-

ally

for

the most

part,

by

a system of

inquiry

units.

Data

required

for

historical

compari-

sons could

be

merged

gradually

into

index or

percentage

figures.

It

is

availability

of

figures

such as

these,

now

in

relatively

mea-

ger

supply throughout

industry,

that

helps/

the

operating

team

to

pull

together

for

the

best

of

the

over-all

enterprise.

Fully

detailed

data

would

be

dumped

periodically

out of

memory

as needed

for

audit,

or

for

permanent

and

minimized

cost,

financial,

and

inventory

records.

Some

Problems

in

Applicable

Equipment and

Methods

The

television

industry

has

concentrated

on

home

receivers.

As

a consequence,

all

pro-

ducers of equipment

are

largely

standardized

for

a

525

line

vidicon

scanner,

which cannot

legibly

transmit

small

print

on

many

current

business forms. This can

be

overcome

by

op-

tical

systems

feeding

smaller

areas

into

the

scanner or

by

redesign

of the form

to

provide

smaller

volume

of

somewhat

larger

print-

in

the

transmission.

Another- tube

called-an

image

orthocon develops

750

lines

for

the

screen,

but

it

costs

about $1,250 compared

to

about $280

for

the

vidicon.

There

is

no

question

that

finer

scanners can

be

engineered

for

higher

prices

and

that

all

prices

will

de-

crease

over a period of time.

Of

course,

the

vidicon should

be

adequate

for

most purposes,

although

finer

equip~nt

is

theoretically

de-

sirable

for,

say,

fine

photo-copy work

if

the

data

system develops around

that

particular

technique.

(cont'd

on page 38)

USE OF A COMPUTER

FOR

CERTAIN

OPERATIONS

OF

CLASSIFICATION

Andrew

D.

Booth

Birkbeck

College

Computational

Laboratory

University

of

London, London, England

It

is

proposed

to

examine,

in

this

paper,

the

way

in

whicn an

automatic

digital

computer

can be

applied

to

the

resolution

of

two

sorts

of

classification

problems. The

first

problem

is

that

of

frequency

analysis

and,

in

particu-

lar,

the

making

of

what

may

be termed a

'con-

cordance'.

The

second problem concerns

the

analysis

of

the

structures

of

groups

of

words

which

may

be

likened

to

sentences

occurring

in

a

text.

At

the

outset

it

should be

made

clear

that

there

is

no

difficulty

in

programming

a computer

to

solve

either

of

the

problems

if

an

adequate

internal

store

is

available.

The

interest

arises

when

the

problems have

to be

solved

with

limited

storage

capacity

and where,

because

they

arise

in

linguistic

applications,

the

number

of

items

to

be

classi-

fied

may

be very

large.

Consider

first

the

problem

of

frequency

analysis.

He~e,

in

linguistic

terms,

it

is

necessary

to

list

the

numbers

of

different

words which

occur

in

a

given

text.

It

is

to

be imagined

that

the

text

to

be

analyzed

is

presented

to

the

machine

in

the

form

of

a

punched

or

magnetic

tape,

and

that

on

this,

each

alphabetic

or

other

symbol

is

repre-

sented

in

a coded form

acceptable

to

the

machine. Each word

is

supposed

to

be

termin-

ated

by

a space symbol

or

by

a

punctuation

mark

whose code can be

recognized

by

the

ma-

chine

as

differing

from

that

which

represents

a

letter.

If

unlimited

internal

storage

ca-

pacity

is

available

the

frequency

analysis

could

proceed

as

follows:

Assume

that,

so

far,

n

different

words

have been

encountered

in

the

text,

and

that

the

code symbols

for

these

have been

stored

in

locations

a + 1, a +

2,

.••

a + n. Each

location

will

be

filled

as:

(Code symbols) (Spare

digits)

(Count number).

The next word

is

now

read from

the

tape,

its

end

being

recognIzed

by

the

presence

of

a

space

or

punctuation

symbol.

An

attempt

is

now

made

to

identify

the

word

with

one

of

the

words

already

held

in

positions

a + 1, a + 2,

•••

a +

n;

-18

-

this

can be most

conveniently

carried

out

by

the

method

of

t

Bracketing',

first

described

by

the

author(I).

If

the

word

is

found

to

have

occurred

previously,

in

location

a + k

say,

unity

is

added

to

the

count number

held

in

that

position.

If,

however,

the

word has not

occurred

previously,

it

is

stored

in

position

a + n + 1 and a count number 1

is

placed

in

the

relevant

count number

position.

This

process

is

continued

until

the

end

of

the

tape

is

reached,at

which

point

the

words and

their

count numbers

are

either

out-

put

directly,

or

sorted

into

alphabetic

or

frequency

order

and then

printed

out.

When

only

limited

internal

storage

is

available

this

simple

process

is

impossible,

but

it

has been found

that

the

follow'ing

variant

gives

satisfactory

results.

It

is

assumed

that

the

available

internal

storage

extends

from a + I

to

a +

L.

(a)

Read

the

from

the

input

tape.

(b) Compare

with

words

already

stored.

(c,l)

If

previously

encountered,

add

unity

to

the

relevant

count

number

and

return

to

(a).

(c~)

If

not

previously

encountered,

test

to

see

if

available

storage

is

filled.

W,D

If

storage

is

available

record

word and

unit

count number

in

location.

Record

filling

of

additional

storage

space

and

return

to

(a).

(d,2)

If

storage

is

full

punch

out,

or

otherwise

record,

the

word

on

out-

put

tape.

Return

to

(a).

It

is

clear

that

the

result

of

this

pro-

cess

will

be an

analysis

of

the

frequency

of

occurrence

of

the

first

L

different

words

to

be used

in

the

text,

and a

new

tape

from which

these

words have been

eliminated.

The

process

is

now

repeated

on

the

derivative

tape

and

the

cycle

repeated

until

all

of

the

text

has been

dealt

with.

The more

general

problem

of

preparing

a

concordance

involves,

not only

the

statistical

Computer

Use

analysis

described

above,

but

also

the

provi-

sion

of

a

list

of

page and

line

numbers

on

which each word

appears.

This

can be handled

in

a

precisely

similar

way

but

involves

the

use

of

several

computer

storage

locations

to

hold

the

data

for

each

text

word.

The

input

tape

is

now

provided

with

page and

line

end

symbols and

these

are

used

to

insert

the

necessary

concordance

data

into

the

storage

locations

associated

with

the

word under

ex-

amination.

The

program

of

uperations

is

as

follows:

(a)

(b)

Read symbol from

tape.

Examine

to

see

if

it

is:

1. Line

start

symbol,

if

so

go

to

(c.l).

2. Page

start

symbol,

if

so

go

to

(c,2).

3.

Alphabetic

symbol,

if

so

store

with

go

to

(a).

4. Space symbol

or

punctuation

mark,

if

so

go

to

(d).

(c,D

Increase

current

line

count number

held

in

store

by

unity,

punch

line

start

symbol

on

output

tape

and

return,

to

(a).

(c,2)

Increase

current

page count

num-

ber

held

in

store

by

unity;

re-

set

line

count

number

to

unity,

punch page:

start

symbol

on

output

tape

and

return

to

(a).

(d) Compare assembled word with words

already

stored,

if

present

go

to

(e,!),

if

absent

to

(e,2).

(e,l)

Increase

frequency count

number

by

unity,

insert

page and

line

count numbers

in

storage,

return

to

(a).

(e,2)

Examine

"storage

occupied" count

number.

If

i~

has not reached

the

permissible

limi~

increase

it

by

unity

and

store

the

new

word, page and

line

numbers and

unity

frequency

count number

in

the

location.

If

the

permissible

limit

has been

reached,

punch

the

word onto

the

output

tape

and

return

to

(a).

It

is

evident

that

the

program

just

given

will

result

in

an

output

which

treats

the

vari-

ous forms

of

any word

(singular,

plural,

etc.)

as

different

entities.

If

it

is

desired

to

avoid

this,

all

that

is

jnecessary

is

to

.pro-

vide

the

machine

with

a stem

dictionary(2)

which

will

enable

the

various

forms

of

a

word

to

be

recognized.

Since,

however,

pro-

grams

of

this

type

are

generally

used

to

make

analyses

preliminary

to

the

construction

of

the

stem-ending

dictionary,

it

is

frequently

more

useful

to

produce

the

full

output

in

alphabetical

order.

The

effect

of

this

is,

in

general,

to

associate

all

forms

derived

from

the

same

stem

in

adjacent

positions,

-19

-

which

is

a

considerable

aid

to

the

ensuing

analysis.

For

the

second

of

the

two

problems

men-

tioned

at

the

outset,

the

following

may

be

taken

as

typical:

It

is

required

to

examine

a

given

text

and

to

produce a

list

of

all

sentences

(or

phrases)

having a

given

structure.

Here

again,

given

unlimited

storage

space,

the

problem can be

solved

in

a

straightforward

man-

ner.

Assume

that

the

machine

holds

in

store

a

dictionary

which

associates

each word

with

a

part-of-speech

or

other

grammatical symbol.

The problem

is

to

recognize

a

given

configura-

tion,

say

(sl,

S2,

..•

sn),

in

th~

input

and

to

list

this

as

and

when

it

occurs.

A simple

program would be:

(a)

Read

and

store

all

words

up

to

the

mark.

(b) Compare words,

in

sequence,

with

dictionary

and

thus

obtain

gram-

matical

indication

numbers.

(c)

Compare

indication

numbers

with

(Sl,

s2,

•..

sn).

If

iden/tical,

out-

put

the

given

sentence

and

return

to

~a);

if

not

return

to

(a)

di-

rectly.

With most modern machines

this

method

of

approach

is

quite

practicable

since

structural

groups

are,

in

generaL

not

very

large.

The

disadvantage

lies

in

the

fact

that

the

pro-

gram

required

to

handle

variable

length

sen-

tences

tends

to

be

complicated.

To

overcome

this

the

following

procedure

has been evolved:

(a) Read

in

the

next word and punch

it

on

to

the

output

tape.

(b) Compare

with

dictionary

to

obtain

grammatical symbol.

(c)

Compare grammatical symbol

with

the

appropriate

configuration

symbol,

sr

say.

Test

to

see

if

sr

is

the

last

unit

of

structure

to

be

considered.

WJ)

If

the

configuration

symbols

are

identical,

and Sr

is

not

the

last

unit

of

structure,

return

to

(a).

~~)

If

symbols

are

not

identical

con-

tinue

reading

and punching

until

the

mark

is

reached;

punch

this

out

and

re-

turn

to

(a).

(d,3)

If

the

symbols

are

identical,

and

Sr

is

the

last

unit

of

structure,

continue

reading

and punching un-

til

the

mark

is

reached.

Punch

this

out

twice

in

succession

and

then

return

to

(a).

The

output

tape

produced by

this

process

will

contain

the·

original

text

marked

in

such

a

way

that

each

structural

unit

of

the

required

type

is

followed

by

a double

punctuation

mark.

This

tape

is

now

fed

to

the

machine backwards

(cont'd

on page 37)

ROBO.TS

AND

A

SHORT

AUTOMATA:

HISTORY'

James

T.

Culbertson

California

State

Polytechnic

College

San

Luis

Obispo,

Calif.

(Part

1

of

this

article

with

numbered

refer-

ences

appeared

in

the

March

issue,

and

consisted

of

the

first

chapter

of

a

forthcoming

book,

"The

Minds

of

Robots:

Behavior

and

Sense

Data

in

Hy-

pothetical

Automata".

The

~hole

bibliography

of

the

book

is

here

included

in

Part

2;

the

numbers

used

in

Part

1

agree

with

the

numbers

here.)

BIBLIOGRAPHY

1.

Adrian,

E.

D. ;

The

Physical

Background

of

Perception;

Oxford

University

Press;

1947

2.

Ashby, W.

R.;

Design

for

a

Brain;

New

York:

Wiley,

1952

3.

Attneave,

Fred;

Some

Information

Aspects

of

Visual

Perception;

Psychological

Review;

Vol. 61,

No.3,

1954

4.

Bacon,

Francis;

The

New

Atlantis

5.

Bakewell,

C.

M.;

Source

Book

in

Ancient

Philosophy,

Scribners,

1907

6.

Bales,

Fried

7.

Bavink,

B.;

The

Natural

Sciences,

Appleton-

Century,

1932

8.

Berkeley,

E.

C.;

Giant

Brains.

New York:

Wiley,

1949

9.

Berkeley,

Bishop;

Principles

of

Human.Know-

ledge,

Sections

1,

2,

4,

6, 8, 23

10.

Boring,

E.

G.;

The

Physical

Dimensions

of

Consciousness;

New York;'

Century,

1933

11.

Boring,

E.

G. ;

Sensation

and

Perception

in

the

History

of

Experimental

Psychology;

New

York;

Appleton-Century

Go.,

1942

12.

Brewster,

Sir

David;

Letters

on

Natural

Magic;

Harper

Brothers;

New

York,

1870

13.

Brunswik,

E.;

The

Conceptual

Framework

of

Psychology;

Chicago:

University

of

Chic-

ago

Press,

1952

14.

Bulletin

of

Mathematical

BiophysiCS;

Edited

by

N.

Rashevsky;

University

of

Chicago

Press;

numerous

articles

especially

since

about

1942

15.

Bullfinch's

Mythology,

Chap.

II

16.

Carnap,

R.;

Logical

Foundations

of

the

Unity

of

Science;

Encyclopaedia

of

Unified

Science;

Vol.

I,

No.1,

Univ.

of

Chicago

Press;

1934

-

20-

17.

Castiglioni,

Arturo;

A

History

of

Medicine

(translated

by

E.

B.

Krumbhaar)Alfred

A.

Knopf; New

York,

1947

18.

Chambers

Miscellany

of

Useful

and

Entertain-

ing

Knowledge, Gould

Kendall

and

Lincoln,

(No

date)

19.

Churchman,

C.

West

20.

Cohen

and

Drabkin:

Source

Book

in

Greek

Science;

McGraw-Hill,

1948 (Note

also

Aulus

Gellius

Attic

Nights

X)

21.

Culbertson,

JT.;

A

Physical

Theory

of

Sensa-

tion;

Journal

of

Philosophy

of

Science;

Vol.

9,

No.2,

April,

1942

22.

Culbertson,

JT.;

Distances

and

Qualities

in

Perceptual

Space,

Journal

of

Psychology,

Jan.,

1945

23.

Culbertson,

J.

T.;

Temporal

Change

in

Perpet-

ual

Space,

Journal

of

Psychology,

Jan.,

1946

24.

Culbertson,

J.

T.

; A

Device

for

Optic

Nerve

Conduction

and

Form

Perception

I;

Bulletin

of

Mathematical

Biophysics,

June,

1948

25.

Culbertson,

J.

T.;

A

Device

for

Optic

Nerve

Conduction

and

Form

Perception

II;

Bulletin

of

Mathematical

Biophysics,

June,

1948

26.

Culbertson,

J.

T.;

Consciousness

and

Behavior;

Wm.

C.

Brown

Co.,

1950

27.

Culbertson,

J.

T.

;

Flicker

and

Fusion

in

a

Hy-

pothetical

Device

for

Light

Intensity

Discrim-

ination;

RAND

paper

RM-744,

Nov.,

1951

28.

Culbertson,

J.

T.;

Hypothetical

Robots

and

the

Problem

of

Neuroeconomy;

RAND

Project

P-296,

April,

1952

29.

Culbertson,

J.

T.;

Memoryless

Robots;

RAND

Paper

P-316,

Aug.,

1952

30.

Culbertson,

J.

T.;

Sense

Data

in

Robots

and

Organisms;

RAND

Paper

P-378,

May, 1953

31.

d'Abro,

A.;

Evolution

of

Scientific

Thought,

Liveright,

1927 .

32.

Descartes

(Letter

from

Gassendi

to

Descartes,

1641,

quoted

by

Descartes

in

Objections

and

Replies;

See

Philosophical

Works

of

Des-

cartes,

trans.

E.

S.

Haldane

and

?,

Cam-

bridge

University

Press,

1912,

v.

2

pp.

201-

02.

Quoted

in

The

Spirit

of

Western

Philos-

ophy,

Stallknecht

and Brumbaugh,

Longman~,

Green

&

Co.,

1950,

p.

274)

33.

Dotterer,

R.

H.;

Philosophy

by

w~

of

the

Sciences.

New

York;

MacMillan,

1929.

p.

197ff.

34.

Dudley,

H.,

and

Tarnoczy,

T.

H.,

"The

Speaking

Machine

of

Wolfgang

von

Kempelen";

Bell

Telephone

System

Technical

Publica-

tions,

Monograph,

1734.

Copyright,

1950,

Acoustical

Society

of

America

35.

Eddington,

A;

Nature

of

the

Physical

World,

New

York,

MacMillan,

1929

36.

Edinburgh

Encyclopedia,

1843

History

37.

Einstein,

A.;

Relativity,

Holt, 1931, p.

65.

38.

Encyclopedia

Universaillustrada,

Madrid,

Article

on

Automata

39.

Feigl,

Herbert

40.

Flood,

Merrill,

Refer

to

RAND

paper

41.

Gassendi;

See

Descartes

42.

Gerard,

R. W. ;

Scientific

Monthly, 62:477, 1946

43.

~rard,

;

Caesar;

Charles

Scribnerrs

Sons,

1952;

Page

544

44.

Gibson,

James

J.;

The

Perception

of

the

Visual

World;

Houghton Mifflin, 1950

45.

Granit,

R.;

Sensory

Mechanisms

of

the

Ret-

ina;

Oxford

Univ.

Press,

London, 1947

46.

Gregory,

J.

C. ;

Short

History

of

Atomism,

A.

and

C.

Black,

London, 1931,

Chap.

5.

47.

Grey-Walter,

;

The

Living

Brain,

W. W.

Nor-

ton

&

Co.,

Inc.,

1953

48.

Hartline,

H.

K.

;

Nerve

Messages

in

the

Fibers

of

the

Visual

Pathway,

Journal

of

the

Optical

Society

of

America,

30, 239-247 (1940)

49.

Hayek,

F.

A.;

The

Sensory

Order;

University

of

Chicago

Press,

Chicago,

1952

50.

Hebb, D.

0.;

The

Organization

of

Behavior;

New

York;

Wiley, 1949

51.

Helmholtz,

H. von;

Physiological

Optics

(trans.

J.

P.

C.

Southall)

Optical

Society

of

America,

1924

52.

Householder,

A. S. ; RAND

paper

RM-671

Neural

Nets

for

Toad

T1,

1951.

RAND

pap-

er

RM-678

Pavlovian

Learning,

1951

53.

Householder,

A. S.

and

Landahl,

H.

D.;

Mathe-

matical

Biophysics

of

the

Qentral

Nervous

System.

Principia

Press,

1945

54. Huxley,

T.

H. ;

Science

and

Culture;

1881

55.

James,

William;

The

Principles

of

Psychology;

New

York:

Henry

Holt

&

Co.,

1890

56.

Kaempffert,

Waldemar,

Explorations

in

Science;

New

York;

The

Viking

Press;

1953,

Page

283

57. Kaplan, A.

58.

Kleene,

S.

C.

;

Representation

of

Events

in

Nerve

Nets

and

Finite

Automata,

RAND

paper

RM-704,

Dec.,

1951

59.

Kluever,

H.;

Behavior

Mechanisms

of

Monk-

eys;

Univ. of

Chicago

Press;

1933

60.

Lashley,

K.

S.;

In

Search

of

the

Engram,

in

"Physiological

Mechanisms

in

Animal

Be-

haVior," New

York,

Academic

Press,

Inc.,

1950

-='

61.

Lashley,

K. S. ;

Functional

Interpretations

of

Anatomic

Patterns,

in

"Patterns

of

Organ-

ization

in

the

Central

Nervous

System",

Philip

Bard,

Editor.

Baltimore;

Williams

and

Wilkins

Co.,

1952

62.

Leibnitz

Mondology

63. Loeb,

J.;

Forced

Movements,

Tropisms

and

Animal

Conduct

64.

Lovejoy, A.

0.;

The

Revolt

Against

Dualism,

-21 -

Open

Court,

1930

65.

Major,

D. R. ;

Introduction

to

Philosophy,

Odyssey

Press,

1933

66.

Manning, H.

P.;

Geometry

of

Four

Dimensions,

MacMillan,

1914

67.

Margenau,

H.;

The

Nature

of

Physical

Reality,

McGraW-Hill,

1950

68.

McCullough, W. S.

and

Pitts,

W.;

A

Logical

Calculus

of

Ideas

Immanent

in

Nervous

Ac-

tivity;

Bulletin

of

Mathematical

Biophysics

5, 1943

68a.

McCullough, W. S.

and

Pitts,

W.;

How We

Know

Universals

69.

McCurdy,

Edward;

The

Mind

of

Leonardo

da

Vinci; Dodd,

Mead,

and

Co.,

1948

70.

McDougall,

William;

Body

and

Mind; New

York;

MacMillan

Co.,

1920

71.

Mead,

H.;

Types

and

Problems

of

Philoso-

phy;

Henry

Holt

&

Co.,

1946

72.

Mettler,

C.

;

History

of

Medicine;

Blake-

ston

Co.,

1947

73.

Miessner,

B.

F.

;

From

Electrical

Experi-

menter,

Sept.,

1915,

p.

202

74.

Northrop,

F.

S.

C.;

Science

and

First

Prin-

ciples,

MacMillan,

1931

75.

Northrop,

F.

S.

C.;

Science,

107:411, 1948

76.

Parry,

T.

W. ;

The

Prehistoric

Trephined

Skulls

of

Great

-Britain;

Proc.

Roy.

Soc.

Med.

(Hist.

Sect.)

Vol. XIV, Suppl,

(Lon-

don, 1921)

77.

Patrick,

G. W. W. ;

Introduction

to

Philosophy;

Houghton Mifflin, 1936

78.

Penfield,

W.

and

Rasmussen,

T.;

The

Cere-

bral

Cortex

of

Man;

MacMillan,

1950

79.

Pirenne,

M.

H.;

Vision

and

the

Eye;

Pilot,

London, 1948

80.

Poe,

E.

A. ;

Maelzelrs

Chess

Player

81.

Polyak,

S.;

The

Retina;

Univ

of

Chicago

Press,

1941

82.

Quastler,

Henry

(editor);

Information

Theory

in

Biology; Univ. of

illinois

Press,

Urbana,

1953

83.

Rapoport,

A

and

Shimbel,

A.;

Cybernetics

and

Mathematical

Biology,

in

ETC.,

a

Re-

view

of

General

Semantics,

Spring,

1949

84.

Rapoport,

A.;

Operational

Philosophy,

Har-

pers,

1953

85.

Rashevsky,

N.;

Mathematical

Biophysics,

1948,

Part

Three

86.

Rosenblueth,

Wiener

and

Bigelow;

Journal

of

the

Philosophy

of

Science,

1943,

10,

18

87.

Russell,

B.;

Human

Knowledge; New

York,

Simon

and

Schuster,

Inc.,

1948

88.

Sarton,

George;

History

of

Science;

Vol. I,

Carnegie

Institute,

1927

89.

Schrodinger,

E.;

Science

and

Humanism;

Cambridge,

1951

90. Shannon, C.

E.

and

Weaver,

W.;

Mathemat-

COmputers

and

Automation

ical

Theory

of

Communication;

Univ.

of

illinois

Press,

1949

91.

Sherrington,

C.;

The

Integrative

Action

of

the

Nervous

System;

Yale

Univ.

Press,

New

Haven,

1920

92.

Sperry,

R.

W.,;

Neurology

and

the

Mind-Brain

Problem;

American

Scientist,

Vol.

40,

No.

2,

April,

1952

93.

Stallknecht

and

Brumbaugh,

the

Spirit

of

Western

Philosophy;

Longmans,

Green

and

Co.,

1950

94. Swift,

J;

Gulliver's

Travels,

Part

ill

95.

Tarski,

Alfred;

Introduction

to

Logic,

New

York,

1941

96.

Titus,

H.

H.;

Living

Issues

in

Philosophy;

American

Book

Co.,

2nd

ed.,

1953

97.

Tolman,

E.

C.;

Purposive

Behavior

in

An-

imals

and

Men; New

York,

Century

1932

98.

Tornebohm,

H.;

A

Logical

Analysis

of

the

Theory

of

Relativity;

Stockholm;

Almquist

and

Wiksell,

1952

99.

Troland,

L.

T.;

The

Principles

of

Psycho-

physiology;

New

York,

Van

Nostrand,

1932

100.

Turing,

A. M. ;

Computing

Machinery

and

Intelligence;

MIND, Oct. 1950

101.

Von

Foerster,

H.

(editor)

Cybernetics,

Cir-

cular

Causal

and

Feedback

Mechanisms

in

Biological

and

Social

Systems;

Transactions

of

the

Sixty

Conference,

1949;

Josiah

Macy

Jr.

Foundation

102.

von

Kempelin

103.

von

Neumann,

J.;

The

General

and

Logical

Theory

of

Automata,

in

"Cerebral

Mechan-

isms

in

Behavior,

Hixon

Symposium,

New

York,

John

Wiley

and

Sons,

1951

104.

von

Neumann,

J.;

Probabilistic

Logics,

Cal-

ifornia

Institute

of

Technology

Lectures,

Jan.,

1952

105.

Wiener,

N.;

Cybernetics,

N.

Y.,

Wiley,

1948

106.

Wiener,

N.;

The

Human

Use

of

Human

Beings,

Houghton-Mifflin,

1950

107.

Willis

108.

Woodger,

J.

H.;

Axiomatic

Method

in

Biology;

Cambridge

Univ.

Press,

1937

109.

Descartes;

Passions

of

the

Soul XXXIV

110.

Creegan,

R.

F.;

(R)-systems

versus

Psycho-

logical

Spaces;

Journal

of

Philosophy

of

Science,

Oct.

,1942

111.

Morgan

and

Steller;

Physiological

Psychol-

ogy,

McGraw-Hill,

1950

112.

Hurewicz,

W.

and

Wallman,

H.,

Dimension

Theory,

Princeton

Vniv.

Press,

1941,

Page

4

113.

Manning, H.

P.;

Geometry

of

Four

Dimen-

sions,

MacMillan,

1914

114.

Barnett,

L.;

The

Universe

and

Dr.

Einstein,

Sloane,

1948,

Chap.

9

115.

Jeans,

J.;

The

Mysterious

Universe,

Cam-

-22 -

bridge

Univ.

Press,

New

Revised

Edition,

1932, pp.

117-121

116.

Korzybski,

A.;

Science

and

Sanity,

2nd

Ed.

,

1941, pp.

626-630

117.

Russell,

B.;

The

ABC of

Relativity,

Keagan

Paul,

1927, p. 58ff.

118.

d'

Abro,

A.;

Evolution

of

Scientific

Thought,

Boni

and

Liveright,

1927,

Chap.

XVllI.

Note

discussion

of

point-

event

and

world-

line

119.

Minkowski,

H.;

In

the

Principle

of

Relativity

by

A.

Einstein

and

Others,

Methuen

&

Co.,

London, 1923,

page

75

120.

Steinmetz,

C.

P.;

Relativity

and

Space,

Mc-

Graw-Hill,

1923, p. 24

121.

Margenau,

H.;

The

Nature

of

Physical

Real-

ity,

McGraw-Hill,

1950, p. 150

122.

Fitch,

F.

B. ;

Physical

Continuity.

Philosophy

of

Science,

3,

No.4,

Oct.,

1936,

486-494

123.

Biser,

I.;

A

general

Schema

for

Natural

Sys-

tems.

Philadelphia,

Westbrook

Publishing

Co.,

1938

124.

Hinton,

C.

H.;

The

Fourth

Dimension.

London: Swann

Sonnenshein

&

Co.,

Ltd.,

1904

125.

Hobbes,

T.;

Leviathan.

London:

J.

M.

Dent,

1924

126.

Lucretius;

De

Rerum

Natura.

English

trans-

lation

by

W. H. D.

Rouse.

London: W.

Heinemann,

1931

127.

Thompson,

D'Arcy

W.;

On

Growth

and

Form.

Cambridge

128.

Culbertson,

J.

T.;

The

place

of

the

sense

world

in

physical

theory;

Ph.

D.

Thesis.

Yale

University,

1940

129.

Bowden,

B.

V.;

Faster

Than

Thought,

Pit-

man

&

Sons,

London, 1953

130.

Lewin,

K.;

Principles

of

Topological

Psy-

chology,

McGraw-Hill,

1936

131.

Gerard,

Ralph

W.;

What

is

memory?

Scien-

tific

American,

Sept.

1953

132.

Fulton,

J.

R.

;

Physiology

of

the

Nervous

System;

2nd

Edition.

New

York:

Oxford

Vniv.

Press.

1943

133.

Watson,

J.

B.;

Behaviorism.

New

York:

Norton

134.

Mead,

Shepherd;

The

Big

Ball

of

Wax;

Simon

&

Schuster;

New

York,

19P4.

p.109ff

135.

McCloy,

Shelby

T.;

French

Inventions

of

the

Eighteenth

Century;

Univ.

of

Kentucky

Press,

1952.

Chap.

VIII

on

automata.

136.

Sinnott,

E.

W.;

Cell

and

Psyche,

Unlv.

of

North

Carolina

Press,

1950

137.

Bonner,

J.

T.

;

Morphogenesis,

Princeton

Vniv.

Press,

1952

138.

Culbertson,

J.

T.

;

Some

Uneconomical

Ro-

bots.

In

Automata

Studies,

Edited

by

History

Shannon, C.

E.,

and

McCarty,

J.

Prince-

ton

Univ.

Press,

1956

139.

Booth, A. D.

and

Booth, K. H.

V.;

Automatic

Digital

Computers;

Butterworths

Scientific

Publications,

1953

140.

Computers

and

Automation.

Monthly

Peri-

odical.

E.

C.

Berkeley

&

Associates

141.

Abbott,

John

A.

Electroencephalography.

Reprinted

from

New

England

Journal

of

Medicine;

Jan.

6 and

Jan.

13,

1955

142.

Report

of

Conlmittee

on

Colorimetry

for

1920-21.

Journal

of

the

Optical

Society

of

America

and

Review

of

Scientific

In-

struments.

Aug. 1922,

§.

p. 535

143.

Bulletin

of

Mathematical

Biophysics.

Ed

N.

Rashevsky

144.

Frankel,

Stanley;

On

the

Design

of

Automata

and

the

Interpretation

of

Cerebral

Behav-

ior.

Psychometrika,

Vol. 20,

Number

2,

June

1955

145.

MacKay,

D. M. ;

Mindlike

Behavior

in

Arte-

facts.

British

Journal

for

the

Philosophy

of

Science,

1951,

II,

P

108

146.

Thrall,

R.

M.,

Coombs,

C.

H.,

and

Davis,

R.

L.;

Editors

Decision

Processes,

Wiley,

New

York,1954

147.

Brain,

W.

Russell;

Mind,

Perception

and

Science;

Blackwell

Scientific

Publications,

Oxford,

1951

148.

Richter,

D.,

Editor;

Perspectives

in

Neuro-

psychiatry;

Lewis,

London, 1950

149.

Von

Newnann,

J.

and

Morgenstern,

0.;

Theory

of

Games

and

Economic

Behavior.

Princeton

University

Press,

1944

150.

Babbage,

H.

P.;

Babbage's

Calculating

Engines.

London:

E.

and

F.N.

Spon.,

1889

151.

Watson,

G.;

"Apparent

Motion

and

the

Mind-

Body

Problem",

British

Journal

for

the

Philosophy

of

Science,

Vol.

II,

No. 7

152.

Shannon,

C.

E.

and

McCarthy,

eds.;

Auto-

mata

Studies,

Princeton

University

Press,

1956

153.

McKay, D.

M.;

"The

Epistemological

Prob-

lem

for

Automata".

In

Automata

Studies,

Edited

by

Shannon,

C.

E.,

and

McCJIthy,

J.,

Princeton

University

Press,

1956

154.

Muller,

D.

E.;

"Complexity

in

Electronic

Switching

Circuits",

I.

R.

E.

Transactions

on

Electronic

Computers,

March,

1956

155.

Minsky,

M.

L.;

"Some

Universal

Elements

for

Finite

Automata".

In

Automata

Studies,

Edited

by

Shannon,

C.

E.

and

McCarthy,

J.,

Princeton

University

Press,

1956

156.

Moore,

E.

F.;

"Gedanken-Experiments

on

Sequential

Machines".

In

Automata

Studies,

Edited

by

Shannon, C.

E.

and

McCarthy,

J.,

-23 -

Princeton

University

Press,

1956

157.

Turing,

A. M.

"On

Computable

Numbers".

Proc.

of

the

London

Mathematical

Society

2,

42

(1936)

158.

McColl,

L.

A. ;

Fundamental

Theory

of

Ser-

vomechanisms.

Van

Nostrand,

1945

159.

Huxley,

A;

Brave

New

World

160.

YOWlg,

J.

Z.;

Doubt

and

Certainty

in

Science.

Clarendon

Press,

1951

161.

Maxwell,

J.

C.;

"On

Governors".

Proc.

Roy.

Soc. 16, 270,

1868

162.

Hazen,

H.

L.;

"Theory

of

Servomechanisms".

J.

Franklin

Institute,

218, 279, 1934

163.

Ahrendt,

W.

R.;

Servomechanism

Practice,

McGraw-Hill,

1954

164.

Richards,

R.

K. ;

Arithmetic

Operations

in

Digital

Computers,

Van

Nostrand,

1955

165.

Pierce,

John

R.;

Book

review,

Scientific

American,

Aug.,

1956

166.

Tsien,

H. S. ;

Engineering

Cybernetics,

Mc-

Graw-Hill,

1954

167.

Diebold,

J.;

Automation,

Van

Nostrand,

1952

168.

West,

J.

C.;

Textbook

of

Servomechanisms,

English

Universities

Press

Ltd.,

1953

169.

Soule,

G.;

Time

for

Living,

Viking

Press,

1955

170.

Cleator,

P.

E.;

The

Robot

Era,

Allen

and

Unwin,

1955

171.

Greene,

J.;

"Man

as

a

servoconiponent",

Control

Engineering,

Oct.,

1954

172.

Tustin,

A;

"Feedback",

Scientific

American,

187:

48-55,

Sept.,

1952

173.

Butler,

Samuel;

Erehwon,

N. Y. ,

Nelson,

1901

174.

Rapoport,

A.;

"Technological

Models

of

the

Nervous

System".

ETC.,

Sept.,

1954

175.

Pierce,

John

R.;

Electrons,

Waves

and

Messages.

Hanover

House,

Garden

City,

New

York,

1956

176.

Berkeley,

E.

C.

and

Wainwright,

L.;

Com-

puters,

Their

Operation

and

Applications,

Reinhold

Publishing

Co.,

New

York,

1956

177.

Aiken, H. H.

and

others;

Description

of

a

Magnetic

Drwn

Calculator.

Harvard

Univ.

Press,

1952

178.

Allport,

R.

H.;

Theories

of

Perception

and

the

Concept

of

Structure.

Wiley, 1955

179.

Allport,

F.

H.;

The

Structuring

of

Events.

Psychol.

Rev.

61, 1954

180.

Adrian,

E.

D.;

Bremer,

F.

and

Jasper,

H.

H.,

Editors;

Brain

Mechanisms

and

Consciousness.

Blackwell

Scientific

Pub-

lications.

Oxford,

1954

181.

Kohler,

W.

and

Held,

R.;

The

Cortical

Correlate

of

Pattern

Vision.

Science,

110,

414-419.

1949

182.

Lorente

de

No,

R.;

Analysis

of

the

Activity

COmputers

and

Automation

of

Chains

of

Internuncial

Neurons.

J.

Neurophysiology

7,

1938

183.

Schrodinger,

E;

What

Is

Life?

New

York.

MacMillan,

1946

184.

Titchener,

E.

B.;

Experimental

Psychology

of

the

Thought

Processes;

MacMillan,

1909

185.

Titchener,

E.

B.;

A Textbook of

Psychology,

MacMillan,

1914

186.

Tolman,

E.

C.

and

Brunswik,

E.;

The

Or-

ganism

and

the

Causal

Texture

of the

En-

vironment.

Psych.

Rev.,

42,

1935

187.

Adrian,

E.

D.;

Rhythmic

Discharges

from

the

Thalamus.

J.

Physiology,

113,

1951

188.

Bartley,

S.

H.;

Vision.

Van

Nostrand,

1941

189.

Pavlov,

I.

P.

;

Lectures

on

Conditioned

Re-

flexes.

Conditioned

Reflexes

and

Psychi-

atry.

Translated

and

edited

by

W.

H.

Gantt.

New

York.

International

Pub-

lishing

Co.,

1942

190.

Laslett,

Peter,

Editor;

The

Physical

Basis

of

Mind.

MacMillan,

1950

191.

Culbertson,

J.

T.;

Mathematics

for

Digital

Devices.

(Text

now

used

in

mimeographed

form

for

two

undergraduate

courses

at

California

State

Polytechnic

College.

To

be

published

for

fall

1957)

192.

Engineering

Research

Associates;

High

Speed

Computing

Devices.

McGraW-Hill,

1950

193.

Keister,

W.,

Ritchie,

A.

E.

and

Washburn,

, S.

H.;

The

DeSign

of

Switching

Circuits,

Van

Nostrand,

1951

194.

"Review

of

Input

and

Output

Equipment

Used

in

Computing

Systems",

Joint

AIEE-

mE-ACM

Computer

Conference,

142

pages

Dec.

1952.

(Published

March

1953)

195.

Proceedings

of

the

mE,

Computer

Issue,

Oct.,

1953

196.

"A

Survey

of

Automatic

Digital

Computers",

Report

111293,

U.

S.

Dept.

of

Commerce,

Office

of

Technical

Services,

Washington

25,

D.C.

197.

Brain,

W.

R.

;

Neurological

Approach

to

Per-

ception.

Philosophy,

1946,

21,

133

198.

Whitehead, A.

N.;

Process

and

Reality,

Mac-

Millan,

1936

199.

Brain,

W.

R.;

Cerebral

Basis

of

Conscious-

ness,

Brain,

1950,

73,

461

,

200.

Physiological

Mechanisms

in

Animal

Be-

havior,

Editors,

Cambridge,

1950

201.

Locke,

J.;

Essay

Concerning

Human

Under-

standing;

Open

Court,

1927

202.

Kant,

I.;

Critique

of

Pure

Reason,

Translated

by

Max

Muller,

MacMillan,

1925

203.

Mach,

E.;

Analysis

of

Sensations,

Translated

by C. M.

Williams,

Open

Court,

1897

204.

Broad,

C.

D.;

Scientific

Thought,

Harcourt,

Brace,

and

Co.,

1927.

Chaps.

8,

9,

10

-

24

-

205.

Haeckel,

E.;

Riddle

of

the

Universe,

Har-

pers,

1902,

Chap.

10

206.

Parish,

E.;

Hallucinations

and

illusions,

Scribners,

1897

207.

Eddington, A.

S.;

Space,

Time

and

Gravita-

tion;

Cambridge

Univ.

Press,

1935,

Chap.

3

208.

Lewis,

G.

N.;

The Anatomy of

Science,

Yale

Univ.

Press,

1929

209.

Ramsperger,

A. G. ;

Philosophies

of

Science,

Crofts,

1942,

Chap. 4

210.

Reiser,

O.

L.;

The

Alchemy

of

Light

and

Color,

Norton,

1928

211.

Hume,

D.;

Treatise

of

Human

Nature,

Ev-

eryman's

Library,

1930

212.

Carnap,

R.;

Scheinprobleme

in

der

Philos-

ophie,

Weltkreis-Verlag,

Berlin,

1928

213.

Poincare,

H.;

Science

and

Hypothesis,

Dover

Publications,

1952

214.

Weyl,

H.;

Space-Time-Matter,

Methuen,

1922

215.

Alexander,

S.;

Space,

Time'.and

Deity,

Mac-

Millan,

1934

:',

216.

Johnson,

S.;

Rasselas,

Oxfor,~

Univ.

Press,

1949,

Chap.

XLVIII

217.

Steward,

Dugald;

Works.

Edited

by

W.

Ham-

ilton,

1858

218.

Sarisberiensis,

J.;

Collected

Works,

Edited

by

Giles,

J.

A.,

Oxford,

~848:.

(John

of

Salisbury's

12th

Century

associ~tional

psychology.

)

219.

Comte,

A.;

Fundamental

Principles

of

Posi-

tive

Philosophy.

Translated

by

P.

Descours

and H. G.

Jones,

London,

1905

220.

Kohler,

W.;

The

Place

of Value

in

a

World

of

Facts,

Live

right,

1938

221.

Plato;

Dialogues.

Trans.

by

B.

Jowett

Five

Vols.,

Oxford Univ.

Press,

1931.

Theaetetus,

pp.

151-154,

156-161,

182

222.

Brillouin,

Leon;

Science

and

Information

Theory,

Academic

Press,

1956

223.

Granit,

R.;

Receptors

and

Sensory

Percep-

tion,

Yale

Univ.

Press,

1955

224.

Attneave,

Fred;

Amer.

Journal

of

Psychology,

68,

69

-82,

1955

225.

von

Bonin,

G.;

Essay

on

the

Cerebral

Cortex,

Springfield,

Ill.,

Charles

Thomas,

Publish-

er,

1950

'-

226.

Cobb,

Stanley;

On

the

Nature

and

Locus

of

.

Mind,

Archives

of

Neurology

and

Psychiatry,

Vol.

67,

1952

227.

Brazier,

M. A. B. ; A

Neuronal

Basis

of

Ideas,

Dialectica

4:

73,

1950

228.

Gerard,

R.

W.;

Brain

Physiology:

A

Basic

Science;

In

Changing

Concepts

of

Psycho-

analytic

Metiicine;

Grune

and

Stratton,

1956

229.

von

Bonin,

G.;

Notes

on

Cortical

Evolution,

Archives

Neurology

and

Psychiatry,

Vol.

67,

Feb.

1952

230.

Penfield,.

W. ;

Memory

Mechanisms,

Ar-

(cont'd

on page

38)

, ,

WESTERN

JOINT

COMPUTER

CONFERENCE,

LO~

ANGELES,

FEB.

26

TO

FEB.

28, 1957,

PROGRAMS,

TITLES,

AND

ABSTRACTS

The

1957 Western

Joint

Computer

Confer-

ence took

place

at

the

Statler

Hotel,

Los

Angeles,

Calif.,

February

26

to

February

28,

1957. Following

is

the

program, sponsored

jointly

by

the

Institute

of

Radio

Engineers,

the

American

Institute

of

Electrical

Engi-

neering,

and

the

Asspciation

for

Computing

Machinery.

All

registrants

receive

a

free

copy

of

the

proceedings.

Others

may

order

a copy

of

the

proceedings

from any

of

the

three

sponsoring

societies,

such

as

the

Association

for

Computing Machinery, 2

East

63

St.,

New

York

21,

N.

Y.

(The

following

numbers

of

papers

were not

in

the

program,

but

have been added,

for

convenience

in

reference,

by

the

editor

of

"Computers and

Automation".)

Tuesday Morning

1.

Welcome,

J.

L.

Barnes,

System Labo-

ratories

Corp.,

Conference Chairman

2. Keynote Address,

J.

M.

Bridges,

Di-

rector

of

Electronics,

Application

Engineer-

ing

Office

of

the

Assistant

Secretary

of

Defense

Tuesday

Afternoon

SESSION

I:

MATHEMATICS

OF

RELIABILITY

Edward Coleman,

UCLA,

Chairman

3.

"Reliability

from a System

Point

of

View",

A.

W.

Boldyreff,

The

Rand

Corp.

--

The

basic

concepts

of

reliability

are

defined

critically

and

the

present

status

of

reli-

ability

of

complex equipment

is

accessed.

The

emphasis

is

on

treating

reliability

as

one

of

the

system

parameters,

all

of

which

have

to

be

properly

balanced

in

optimum

sys-

tem

designs.

4.

"Design

of

Experiments

for

Evaluating

Reliability",

J.

F. Hoffmann, System

Labora-

tories

Corp.

--

Some

of

the

basic

philosophy

of

experimentation

is

discussed,

and

the

role

of

statistics

in

the

analysis

and

interpreta-

tion

of

experimental

data

is

outlined.

A

hypothetical

experiment

reveals

the

methods

-25 -

by

which

statisticians

organize

experimenta-

tion

in

order

to

simplify

analysis

and maxi-

mize

information

obtained.

Completely randomized, randomized

blocks,

Latin

and

Greco-Latin

square

designs

and

their

associated

analyses

for

single

factor

experi-

ments

are

discussed

briefly.

The

organization

of

multiple

factor

experiments

is

illustrated

by

the

discussion

of

a

factorial

experiment

involving

use

of

split

plots.

The

role

and

necessity

of

randomization

is

emphasized.

5.

"Reliability

and Computers",

W.

H.

Ware,

The

Rand

Corp.

--

The

problem

of

reli-

ability

in

a computing system

is

examined,

and

the

contrast

between

analog

and

digital

systems

in

this

respect

is

discussed.

The

application

of

the

two

to

providing

reliable

computing system

operation

is

developed.

SESSION

II:

NEW

SYSTEMS

A

Robert

Johnson, General

Electric

Co.,

Chairman

6.

"A

Digital

System

Simulator",

W.

E.

Smith,

Aeronutronic

Systems,

Inc.

--

The

digi-

tal

systems

simulator

is

a

device

for

simulat-

ing any

digital

system which

is

capable

of

being

represented

by

flip-flops

and

and/or

decision

elements,

or

more

generally

speaking

by

memory

and

logic.

No

physical

changes

are

required

in

causing

this

simulator

to

assume

the

characteristics

of

various

systems.

All

changes

are

done

by

coding

or

programming

into

memory.

Extensions

to

the

capacity

of

the

device

for

representing

systems beyond

a

certain

complexity

involve

minor

physical

changes and an

increase

in

the

memory

storage

space.

7.

"A

New

Input-Output

Selection

System

for

Florida

Automatic Computer

(FLAC)

" ,

C.

F.

Summer,

RCA

Missile

Test

Project

--

Details

of

a

new

high speed

input-output

selection

system

are

presented.

The

basic

computer

characteristics

and

the

construction

of

the

computer word

are

described.

This

informa-

tion

is

then

factors

which

dictate

the

design

of

the

logic

necessary

to

control

the

selection

matrix.

Characteristics

are

Computers and

Automation

also

presented

on

a

new

high speed

reliable

relay

with

replacement

hermetically

sealed

contacts.

Finally,

illustrations

are

included

which

cover

the

logic

and

selection

matrix,

the

new

relay

package,

finished

chassis

and

an

over-all

view

of

the

ultimate

computer

lay-

out.

8.

"The

IBM

650

RAMAC

System -Disk

Stor-

age

Operation",

D.

Royse,

IBM

-

This

paper

describes

the

operation

of

a

single-step

data

processing

system. Elements

of

the

system

are

a

flexible

medium-speed

stored-program

compute4

and a

six

to

twenty-four-million-digit

memory

composed

of

one

or

more

magnetic-disk

storage-

arrays.

The

paper reviews

the

basic

computer

and

the

disk-storage

units.

The

computer's

control

of

seek,

'read,

and wri

te

memory

opera-

tions

is

described

with

emphasis

on

speed and

reliability

considerations.

9. "The

IBM

650

RAMAC

System -

Inquiry

Station

Operation",

H.

A.

Reitfort,

IBM

-

The

Inquiry

Station

provides

quick

access

to

the

data

processing

system from remote

locations.

The system

consists

of

a

transmitting-receiving

typewriter

operating

through

a

control

unit

which

synchronizes

the

Inquiry

Station

and

the

IBM

650

RAMAC.

Typed

information

under

control

of

a format

tape

in

the

typewriter

is

stored

on

the

magnetic

drum

for

processing

in

the

computer. The

results

are

sent

back

to

the

typewriter

and

automatically

typed under

con-

trol

of

the

format

tape.

The system has

flexi-

bility

and

provides

quick

access

to

any

or

all

records

stored

in

.the

RAMAC.

Wednesday Morning

SESSION

III:

NEW

COMPONENTS

Cornelius

Leondes,

UCLA,

Chairman

10.

"An

RCA

High-Performance Tape

Trans-

port

System", S. Baybick and

R.

E.

Montijo,

RCA

--

A

high-performance,

multi-channel

digi-

tal

tape

transport

was

developed

to

meet

the

needs

of

the

data

processing

industry

in

gener-

al.

This

is

a

tubeless

equipment

using

semi-

conductor

and magnetic components.

This

paper

describes

the

electronics

and

mechanism

in

detail,

including

the

methods

employed

in

obtaining

start-stop

rates

to

120

per

second,

start

and

stop

times

of

less

than

12

milliseconds,

and a

start-stop

spacing

of

less

than

0.2

inches.

The

transport

handles

various

widths

of

tape

from

1/2"

to

1

1/8"

and magnetic heads which

provide

up

to

18

recording

tracks.

11.

"A

Medium

Speed Magnetic Core

Memory",

G.

E.

Valenty,

Remington Rand Univac

--

The

completely

transistorized

type

S3

magnetic

core

memory

designed

and

built

by Remington

Rand Univac

for

the

Transac

SlOOO

computer

~ses

1145

transistors,

consumes 300

watts,

-

26

-

and

occupies

4

cubic

feet.

A

timing

device

utilizing

magnetic

switch

cores

has been

de-

veloped

to

sequence

the

memory

operation.

A

memory

cycle

requires

20

microseconds and

the

word

is

available

after

6 microseconds.

The

logical

circuitry

consists

of

diode

"and" and

"or"

circuitry

plus

transistor

amplifiers.

Low

and high frequency

transistors

are

em-

ployed

to

generate

well

regulated

constant

current

pulses

to

drive

the

memory.

12.

"Millimicrosecond

'FTansistor

Switch-

ing

Techniques",

E.

J.

Slobodzinski

and

H.

S.

Yourke,

IBM

Research

Center

--

A program

was

initiated

to

develop semiconductor

circuitry

capable

of

performing

five

sequential

logical

operations

in

100

milliseconds.

The advent

of

the

drift

transistor

and improved

current

switching

techniques

have

made

these

goals

feasible.

The

design

philosophy

that

resulted

from

this

investigation

will

be

discussed.

Circuits

will

be

shown

that

are

not

critical

with

respect

to

variations

in

alpha

and

Ico.

The

driving

capabilities

and

switching

speeds

of

these

circuits

will

be

discussed.

13. "The

Utilization

of

Magnetic

Domain

Wall

Viscosity

in

Data Handling

Devices",

V.

L.

Newhouse,

RCA

--

The

investigation

of

the

switching

behavior

of

metal

tape

rectangu-

lar

loop

cores

in

the

millimicrosecond

region

has

led

to

the

discovery

of

a group

of

phe-

nomena

collectively

referred

to

as

the

mag-

netic

inertia

effects.

Various

digital

circuit

applications

are

described.

These

include

a

technique

of

con-

tinuously

displaying

the

contents

of

magnetic

shift

registers

and

increasing

the

spe~d

of

operating

random access memories

without

an

increase

in

the

amount

of

equipment

required,

and

without

the

use

of

extra

windings

or

of

a

special

core

geometry.

SESSION

IV:

APPLICATIONS

OF

RELIABILITY

PRINCIPLES

Gilbert

D.

McCann,

Calif.

Inst.

Tech, Chairman

14.

"Reliability

in

Business

Systems",

H.

T.

Glantz,

J. Diebold &

Associates,

Inc.

--

Although

general

agreement

exists

that

scientific

and commercial

data

processing

systems

are

different,

the

precise

nature

of

these

variations,

with

their

res~lting

impli-

cations,

has not

yet

been

clearly

defined.

One

reason

for

this

is

that

business

data

processing

systems

are

used

for

such a wide

variety

of

applications,

with

each system

operating

under

the

restraint

of

a number

of

unrelated

but

dogmatic

outside

agencies

(I.C.C.,

S.E.C~,

Bureau

of

Internal

Revenue,

etc.);

utilizing

data

that

is

frequently

un-

controllable

as

to

format,

scheduling,

or

accuracy;

and

facing

intractable

time

dead-

lines.

This

paper

sets

forth

a method

of

Western

Computer

Conference

approach

that

has

proven

useful

in

determining

the

reliability

requirements

of

systems

that

operate

under such

hazardous

conditions.

15.

"On

Prediction

of

System

Performsnce

from

Information

on

Component

Performance",

~.

R.

Rosenblatt,

National

Bureau

of

Standards

--

This

paper

proposes

some

building

blocks

for

an approach

to

useful

representations

of

system

performance

or

reliability

as

a

func-

tion

of

pertinent

aspects

of

component

per-

formance.

Particular

attention

is

given

to

the

use

of

information

on

ways

in

which com-

ponents

are

believed

or

known

to

be

inter-

dependent.

Some

simple

hypothetical

examples

are

given

of

mathematical

expressions

for

the

relation

of

system

behavior

to

the

behavior

of

interdependent

components.

Consideration

is

given

to

estimation

and

prediction

of

sys-

tem

reliability

using

data

on components, and

to

use

of

the

mathematical

expression

to

simu-

late

the

effect

of

variations

of

component

behavior

on

system

performance.

16.

"Evaluation

of

Failure

Data",

H.

I.

Zagor,

American Bosch

Arma

Corp.

--

The

evalu-

ation

of

failure

data

can be

made

if

graphs

of

failures

vs.

operating

time

are

plotted

and

the

data

tested

for

adherence

to

(1)

Poisson,

and (2)

negative

binomial

distributions.

Methods

of

cumulative

probability

can be

em-

ployed

to

select

items..

An

example

will

be

given

of

an

analysis

of

amplifier

failures

in

requiring

immediate

reliability

attention.

An

example

will

be

given

of

an

analysis

of

amplifier

failures

in

a computing

laboratory.

The

calculations

and

formulas

derived

will

be

shown, and

their

application

to

electronic

equipments

to

(1)

analyze

high

unreliability

items,

(2)

estimate

spares,

(3) compare com-

petitive

equipments,

and (4)

set

up

logistics

procedures

will

be

discussed.

Wednesday Luncheon

17. "Computers

with

European

Accents",

Arthur

L.

Samuel,

IBM

Wednesday

Afternoon

SESSION

V:

COMPONENT

RELIABILITY

Ronald Cone, North American

Aviation,

Inc.,

Chairman

18. "Accuracy

Control

Systems

for

Mag-

netic-Core

Memories",

A.

Katz,

A.

G.

Jones,

G.

Rezek,

RCA

--

Because

of

its

simplicity

and

reliability,

the

coincident-current

mag-

netic-core

memory

has become

the

standard

storage

medium

in

high-speed

digital

com-

puters.

The

present

paper

is

concerned

with

means

for

the

immediate

detection

and speedy

location

of

faults

in

the

memory

system.

Two

such means

are

described:

one,

for

the

-27 -

register

selection

channels;

the

other,

for

the

information

channels.

In

association

with

appropriate

alarm

logic

in

the

control

element

of

the

computer,

these·

means complement each

other

in

enhancing

overall

system

reliability.

19.

"Design

of

Basic

Computer

Building

Block",

J.

Alman, P.

Phipps,

D.

Wilson,

Rem-

ington

Rand Univac

--

This

paper

describes

a

system

of

circuit

design

which

uses

a high

speed

digital

computer

to

do

most

of

the

de-

sign

work. The

digital

computer

will

opti-

mize

the

circuit

by

commutating component

specifications

and

checking

the

circuit

out-

put

requirements;

therefore,

allowing

the

circuit

designer

to

choose

a

circuit

that

will

operate

with

the

widest

tolerance

on

individual

components.

After

this

circuit

is

chosen

the

digital

computer

is

again

called

upon

to

check

this

optimum

circuit.

The com-

puter

does

this

by

calculating

the

variations

in

components

that

will

make

the

circuit

fail

to

perform.

20.

"Error

Detection

in

Redundant

Sys-

tems", S.

Schneider

and

D.

H.

Wagner,

Bur-

roughs Corp.

--

The

paper

is

addressed

gen-

erally

to

the

problem

of

automatic

detection

of

error

of

switching

to

a

redundant

stand-

by. Comparator methods

to

pass

a

majority

signal

from a

triplicated

system

are

pre-

sented

and

evaluated.

Some

techniques

are

discussed

for

low-level

duplication

and low-

level

preventive

maintenance--these

are

less

promising

because

of

the

difficulty

of

auto-

matic

error-deteotion.

SESSION

VI:

ANALOG

COMPUTER

EQUIPMENT

Hans

Meissenger,

Hughes

Aircraft

Co.,

Chairman

21. "Analog

Logari

thmic and

Antilogari

th-

mic

Circuits

Using

Switch

Transistors",

A.

J.

Schiewe and

K.

Chen, Westinghouse

Elect.

Corp.

--

Analog

logarithmic

and

antilogarithmic

cir-

cuits

are

described

which

take

advantage

of

the

inherent

reliability

of

junction

transis-

tors

and

their

stability

when

operated

in

the

switching

mode.

The

operation

of

the

circuits

is

based

upon

the

use

of

exponential

time

de-

cays

in

conj unc t

ion

wi

th

pul

se-width

modulation.

Circuitry

is

described

for

performing

algebraic

operations

on

the

P.

W.

M.

Signals

(add.,

subt.,

mult.,

and dive

by

a

constant).

This

algebraic

circuitry

uses

a

magnetic

core

of

constant

volt-time

area.

Test

results

in

using

the

log-algebraic-antilog

computer

en-

semble

as

a

multiplier

and

as

a

square-rooter

are

given.

22. "High

Precision

Digital-to-Analog

Conversion

by

Integration

of

a

Variable

Rate

Pulse

Train",

A.

D.

Glick,

Minneapolis-Honey-

well

Regulator

Co.

--

A method

of

converting

from

straight

binary

to

an

analog

voltage

or.

Computer'S and

Automation

shaft

position

is

described.

The

analog

out-

put

is

produced

by

integrating

a

train

of

standard

pulses

whose

effective

rate

is

de-

pendent

on

the

binary

input.

The

system,

designed

for

an

airborne

digital

computer,

provides

a continuous analog

output

for

each

binary

input

and responds

in

less

than

one

microsecond

to

a change fn

the

input.

The

results

of

tests

performed

on

the

system

in-

dicate

a

conversion

precision

of

one

part

in

ten

thousand.

23.

"A

Reliable

Method

of

Drift

Stabil-

ization

and

Error

Detection

in

Large-Scale

Analog Computers",

E.

Eddy, Goodyear

Air-

craft

Corp.

--

This method

'of

drift

stabil-

ization

and

error

detection

makes

use

of

a

multi-channel

mechanical commutator and single

stabilizer

amplifier

to

stabilize

many

d.c.

amplifiers.

Past

experience

has

shown

the

following

to

have major

influence

on

system

reliability:

switch

leakage, switch

phasing,

mechanical

failure,

capacitive

couplings

and

hum

pickup. Improvements

in

the

commutator

and

the

use

of

new

and novel

circuitry

have

greatly

increased

the

reliability

of

this

system. This

provides

a convenient

central

point

for

the

detection

of

faulty

operation

of

the

computer.

An

alarm system using

three

separate

fault

indicators

pin-points

defective

units.

24.

"A

New

Method

of

Verifying

analog

Computer Problems and Performance",

W.

C.

Meilander, Goodyear

Aircraft

Corp.

--

Elec-

troni~

differential

analyzers

have become

ex-

tremely

useful

tools

in

research

and

develop-

ment

work

during

the

past

decade. A

factor

limiting

the

extended

utilization

of

analog

computers has been

the

question

of

reliabilit~

not only

of

computer performance,

but

of

oper-

ator

performance as

well.

This paper

describes

methods

of

verifying

that

an

analog computer

problem has been

properly

wired,

that

proper

scale

factors

have been chosen,

that

"in

use"

computer components

are

functioning

properly

and

that

undesired

patching

has not been

made.

Other

desirable

features

in

computer

reliability

are

discussed.

Thursday Morning

SESSION

VII:

NEW

SYSTEMS

B

Montgomery

Pfi~ter,

Ramo-Wooldridge

Corp.,

Chairman

25. "The

Lincoln

TX-2

Computer Develop-

ment",

W.

A.

Clark,

Lincoln

Laboratory,

MIT

--

The

Lincoln

TX-2

incorporates

several

new

developments

in

high-speed

transistor

circuits,

large

capacity

magnetic-core

memories, and

flexibility

in

machine

organization

and

is

designed

to

work

efficiently

with

many

input-

output

devices

of

different

types.

Lincoln

has

constructed

a small

self-checking

multi-

-28 -

plier

system whfch

is

on

life

test,

and a

com-

plete,

though

skeletal,

general-purpose

com-

puter

known

as

the

TX-O

which

is

now

in

operation.

26.

"Functional

Description

of

the

Lincoln

TX-2

Computer",

J.

M.

Frankovitch

and

H.

P.

Peterson,

Lincoln

Laboratory,

MIT

--

The

Lincoln

TX-2

computer

is

a

general-purpose

parallel

binary

machine with a code

of

64

single-address

instructions

and

64

index

registers.

It

includes

a random

access

mem-

ory

of

70,000

36-bit

words with a speed

of

160,000

36-bit

additions

per

second. A unique

feature

of

the

central

computer

is

its

ability

to

deal

with operands

in

one

36-bit,

one

27

and one

9-bit,

two

IS-bit,

or

in

four

9-bit

configurations.

-

27. "The

Lincoln

TX-2

Input-Output

Sys-

tem",

J.

W.

Forgie,

Lincoln

Laboratory,

MIT

--

The

design

utilizes

the

multiple-sequence

program

technique

to

permit

the

concurrent

operation

of

a

number

of

input-output

devices.

A

stored

program

(instruction)

counter

is

associated

with each

input-output

device.

A

priority

system ranks

the

devices

according

to

speed and type

of

efficient

operation.

The

multiple-sequence

program

technique

provides

an

environment

in

which

buffer

storage

may

be

reduced.

2S.

"Memory

Units

of

the

Lincoln

TX-2",

R.

L.

Best,

Lincoln

Laboratory,

MIT

--

Two

of

thesa

core

memories

are

used

for

conventional

storage

of

data

and

instructions.

The

third

is

used as a

file

of

index

registers

and

pro-

gram

counters.

The

largest

memory

contains

65,536 words 37

digits

long.

The

is

entirely

transistor

driven

and

contains

4,096 words 37

digits

long.

The

smallest

and

fastest

contains

64

words

19

digits

long and

uses

external

selection

and

two

cores

per

bit

to

achieve

a

"read"

cycle

time

of

1 usec and

a

"write"

cycle

time

of

3 usec.

29.

"Standardized

Circuitry

for

the

Lincoln

TX-2",

K.

H.

Olsen,

Lincoln

Labora-

tory,

MIT

--

Only

two

basic

transistor

logic

circuits

are

used

in

TX-2,

surface

barrier

transistors

in

saturated

emitter-followers

and

saturated

inverters.

Circuit

tolerance

to

variations

in

tran-

sistor

and

other

component

characteristics,

in

temperature,

supply

voltages,

and

noise

was

studied.

The

study

led

to

the

selection

of

voltage

sensitive

parameters

for

indicat-

ing

the

deterioration

of

components with age

and became

the

basis

of

the

marginal checking

system.

Western

Computer

Conference

SESSION

VIII:

PROGRAMMING

FOR

RELIABILITY

Gene

Amdahl,

Aeronutronic

Systems,

Inc.,

Chairman

30.

"Diagnostic

Techniques Improve

Reli-

ability",

M.

Grems,

R.

K.

Smith,

~~.

Stadler,

Boeing

Airplane.Co.

--

Diagnostic

techniques

are

aids

to

testing,

sampling and

spot-check-

ing a computer program.

Reliability

is

an

accumulation

of

confidence,

a~surance,

and

trustworthiness

promoted

by

these

techniques.

Three

levels

of

'routines

employing

these

tech-

niques

are

discussed.

Examples

are

given

of

routines

which

detect

errors

and

stop

the

machine, which

detect

errors

and

record

perti-

nent

data

before

stopping,

and which

detect

errors,

record

data,

take

corrective

measures

and

then

continue.

Advanced

diagnostic

tech-

niques

not only improve

reliability

but

are

essential

to

computing systems.

31.

"Error

Detection

and

Error

Correction

in

Real

Time

Digital

Computers",

A.

Ralston,

Bell

Telephone

Laboratories

--

A number

of

programming

techniques

are

presented

for

the

detection

and

correction

of

random

transient

computer

errors

in

a

digital

computer

operat-

ing

in

a

real

time environment. The term

correction

is

meant

to

signify

obtaining

either

the

true

value

of

the

quantity

in

error

or

a

sufficiently

good

approximation

to

enable

the

real

time system

to

continue

operation.

Com-

puter

errors

are

classified

according

to

the

seriousness

of

the

error

to

the

system

opera-

tion

and

the

techniques

presented

are

this

classification

of

errors.

The

tech-

niques

include

application

of

well-known mathe-

matical

methods

as

well

as

the

use

of

newer,

more

specialized

methods.

32. "The

FORTRAN

Automatic Coding System,

Description

and

Users'

Reports",

J.

W.

Backus,

et

aI,

IBM

--

The

FORTRAN

Automatic Coding

System

enables

the

programmer

to

communicate

with

the

IBM

704

using

a language

considerably

more

familiar

and

concise

than

the

language

of

machine

instructions.

A program

of

about

22,000

instructions

enables

the

704

to

accept

a

FORTRAN-language

program and produce from

it

a 704-language program, ready

to

run.

The

FORTRAN

language

is

intended

to

be

capable

of

conveniently

expressing

most

pro-

cedures

of

numerical

computation.

Much

of

the

translation

procedure

is

devoted

to

producing

a machine-language program which

will

run

at

about

the

same

speed

as

one

written

by

a good

programmer.

Following a

description

of

the

FORTRAN

System,

members

of

several

computing

installa-

tions

will

describe

their

experiences

in

using

it.

-

-29

-

SESSION

IX:

SYSTEMS

RELIABILITY

J.

Howard

Parsons,

Hughes

Aircraft

Co.,

Chairman

33. "The

Interpretation

and

Attainment

of

Reliability

in

Industrial

Data Systems",

B.

K.

Smith,

Beckman

Instruments',

Inc.

--

No

compon-

ent

is

unreliable

except

as

made

so

the

way

it

is

used. The

practicality

of

proper

use should

determine

the

choice

of

component,

rather

than

life

statistics

taken

in

any

arbitrary

environ-

ment.

Fortunately,

many

of

the

ways

to

reli-

ability

also

result

in

design

and

production

economies. The

inclusion

of

a

digital

com-

puter

as a

part

of

an

industrial

control

sys-

tem

can be a

complication

useful

to

the

guaran-

tee

of

continuous

system

operation.

The

built-

in

intelligence

may

be used

to

race

system

inertia,

and

through

rapid

repair

make

'the

difference

between momentary

failure

and

com-

plete

breakdown.

34. "Accuracy

Control

in

the

RCA

Bizmac

System",

1.

Cohen,

J.

G.

Smith,

A.

1'11.

Spielberg*,

RCA,

(*Formerly

with

RCA,

now

with

General

Electric)

--

The

RCA

Bizmac System has

de-

veloped a

philosophy

designed

to

insure

the

maintenance

of

accuracy

and

reliability

in

its

overall

data

processing

functions.

In

its

system

design

emphasis

is

placed

on

the

accuracy

requirements

of

all

machines

acting

as

parts

of

an

integrated

data

processing

system as

well

as

upon

the

individual

machine

requirements.

Adequate checks

are

provided

in

all

machines

of

the

system

to

maintain

reliable

operation,

and a

further

set

of

checks

is

provided

for

the

system

as

a whole

to

maintain

reliability

in

data

transfer.

The

basic

design

of

the

system

controls

was

engineered

to

incorporate

such

checks.

35. "Continuous Computer

Operational

Reliability",

R.

D.

Briskman,

Army

Security

Agency

--

The

paper

is

a

study

of

system

re-

quirements

to

approach

maximum

reliability

in

the

operation

of

the

computer which must

func-

tion

continuously.

Various

type

complexes,

composed

of

multiple

computers,

are

discussed

in

relation

to

system

down

time,

allowable

output

error

level,

computer

costs,

continuity

in

data

output,

and

probability

of

failure.

36.

"Field

Performance

of

a

New

Automatic

Fault

Locating

Means",

J.

F.

Scully,

Monroe

Calculating

Machine

Co.

and

L.

P.

Colangelo,

Rome

Air

Development

Center

--

The

shortage

of

adequately

trained

personnel

in

the

Air

Force has

not

only

hampered

research

and

de-

velopment programs,

but

has

adversely

affected

sound

field

maintenance

and

the

reliability

of

electronic

equipment under

operational

(cont'd

on page 35)

ASSOCIATION

FOR

CO~PUTING

MACHINERY,

LOS

ANGELES

CHAPTER

MEETING,

LOS

ANGELES,

MARCH

1,

1957

"NEW

COMPUTERS:

A

REPORT

FROM

THE

MANUFACTURERS"

Immediately

following

the

Western

Joint

Computer Conference

in

Los Angeles,

February

26

to

28,

the

Los Angeles

Chapter

of

the

Association

for

Computing Machinery

held

a

Symposium,

"New

Computers: A

Report

from

the

Manufacturers".

Following

is

the

substance

of

the

pro-

gram, which

did

not

contain

abstracts.

In

the

past

technical

papers

representing

computer systems have been

presented

at

tech-

nical

meetings such

as

the

Joint

Computer Con-

ferences

and

meetings

of

the

Association

for

Computing

Machi~ery.

The

question

often

aros~

"Is

a

paper

given

by

a computer

manufacturer

describing

a

new

computer system a

true

and

proper

technical

research

paper

for

such a

conference?"

Two

thoughts

seem

to

run

in

opposition:

first,

the

description

of

a

new

computer system

by

a

manufacturer

is

often

an

advertising

venture

and not

proper

for

a

technical

meeting;

and, second,

public

pre-

sentations

of

detailed

technical

aspects

of

new

computer systems

are

an

important

means

of

communicating

information

necessary

and

important

to

the

user.

It

seems

clear,

however,

that

it

is

en-

tirely

proper

for,

if

not

the

responsibility

of,

the

professional

computer

societies

to

provide

a forum

for

the

manufacturers

to

pre-

sent

and

discuss

new

computers;

the

stated

purpose

of

these

groups

is

to

exchange and

disseminate

technical

information

in

the

com-

puter

field.

In

recognition

of

this

and

the

increased

needs

for

information

exchange

in

this

rapidly

developing

field,

the

Los

Angeles

Chapter

of

the

Association

for

Computing

Ma-

chinery

is

sponsoring

this

Symposium.

The Symposium

is

one

of

the

first

intended

solely

for

the

presentation

and

discussion

of

technical

details

of

new

computer systems.

Because

the

field

has grown so

large,

the

pro-

gram has been

restricted

to

papers

describing

general-purpose,

large-scale

systems

for

scien-

tific

and

b~siness

applications.

Furthermore,

it

was

felt

that

the

Symposium would

serve

the

best

interests

of

users

and

potential

users

everywhere

if

only

the

newest, most advanced,

and most

recently

publicized

computers were

discussed.

Friday,

March

1,

1957

Opening Remarks,

Walter

F. Bauer, The

Ramo-Wooldridge

Corp.;

Chairman,

Los

Angeles

Chapter,

Association

for

Computing Machinery

Session

I

Paul

Armer, The

RAND

Corp.,

Chairman

"Magnetic Tape

File

Processing

with

the

NCR-304,

a

New

Business

Computer",

J.

S.

Sumner,

National

Cash

Register,

Inc.

"The

Cardatron

and

the

Datafile

in

the

DATATRON

System", F.

G.

Withington and

Dean

H.

Shaw,

ElectroData

Corp.

"A

New

Large-Scale

Data Handling System

--DATAmatic 1000",

W.

C.

Carter,

DATAmatic

Corp.

"The

RCA

BIZMAC

II--Characteristics

and

Applications",

J.

A.

Brustman,

H.

M.

Elliott

and

A.

S.

Kranzley,

RCA

Session

II

Jack

A.

Strong,

North American

Aviation,

Inc.,

Chairman

"Advanced Techniques

in

Univac

Scientific

Computer

Systems",

A. A.

Cohen, Remington Rand

Univac \

"Recent

IBM

Developments

in

High Speed

Computation and Design

Objectives

for

the

Super

Speed

Stretch

Computer",

J.

L.

Greenstadt

and

S.

W.

Dunwell,

International

Business

Machines

Corp.

"The

Philco

S-2000

Transistorized

Large-

Scale

Data

Processing

System", S.

Y.

Wong,

Philco

Corp.

"The

Logistics

Research

Model

800

Com-

puter",

Neil

Block,

Logistics

Research,

Inc.

Closing

Rematks, John

W.

Carr

III,

Uni-

versity

of

Michigan;

President,

Association

for

Computing Machinery

-30 -

SYMPOSIUM

ON

SYSTEMS

FOR

INFOR~ATIO~

RETRIEVAL,

WESTERN

RESERVE

UNIVERS

IT'Y

SCHOOL

OF

LIBRARY

SCIENCE,

CLEVELAND,

OHIO,

APRIL

15-17,

1957,

PROGRAM

A Symposium

on

Systems

for

Information

Re-

trieval

is

to

be

held

Monday,

April

15

to

Wednes-

day,

April

17, 1957,

at

Western

Reserve

Univer-

sity,

Cleveland

15, Ohio.

The

host

organization

is

the

School

of

Li-

brary

Science

and

its

Center

for

Documentation

Research.

The

co-sponsors

are

The

Council

on

Documentation

Research,

and

16

other

organiza-

tions

representing

diverse

interests

ranging

from

the

American Bar

Foundation

to

the

Special

Li-

braries

Association.

Following

is

the

program. (The numbers have

been

supplied

by

the

editor

of

"Computers and

Automation". )

Monday,

April

15

Fundamentals

in

Systems Design

1. A Semantic Approach: "Problems

in

Defin-

ing",

PhIlip

B.

Gove,

General

Editor,

G.

&

C.

Merriam

Company

2.

"Classification,

Cataloging,

and

Indexing

Systems", Maurice F. Tauber, Melvil

Dewey

Profes-

sor

of

Library

Service,

School

of

Library

Service,

Columbia

University

3.

"Organizational

Problems

of

Technical

Ab-

stracting

in

the

Field

of

Applied

Mechanics",

Stephen

Juhasz,

Executive

Editor,

Applied

Mechan-

ics

Reviews, Southwest

Research

Institute

4.

An

Operations

Research

Approach:

"Report

of

a Study

for

the

National

Science

Foundation",

Russell

L.

Ackoff

(Director),

Joseph McCloskey,

Operations

Research

Group, Case

Institute

of

Technology

5.

An

Engineering

Approach: "The

Basis

for

a General

Th~ory

of

Documentation", James

W.

Perry

(Director),

Center

for

Documentation and

Communi-

cation

Research', School

of

Library

Science,

Western

Reserve

University

Semi-Automatic Systems

6.

Hand-sorted

Punched Cards -

Subject

field,

"Metallurgy"

-

Presented

by

Marjorie

R.

Hyslop,

Managing

Editor,

Metal

Progress;

Betty

Bryan,

Associate

Editor,

Metals Review, American

Society

for

Metals;

Thomas

H.

Rees,

Research

Assistant,

-

31

-

Center

for

Documentation and Communication Re-

search,

School

of

Library

Science,

Western Re-

serve

University

7.

Machine-sorted

Notched Cards:

"Experience

in

Setting

Up

and Using

the

Zatocoding System" -

Subject

field,

"Aeronautical

sciences"

-

Presented

by

Claude

W.

Brenner,

Allied

Research

Associates,

Inc.

8.

Uniterm Cards -

Subject

field,

"Water

pollution"

-

Presented

by

Patricia

Mines, Case

Institute

of

Technology

9. The Peek-a-Boo System

(Batten-Cordonnier)

-

Subject

field,

"Instrumentation"

-

Presented

by

W.

A.

Wildhack, Joshua

Stern,

Office

of

Basic

Instrumentation,

National

Bureau

of

Standards

10. "The Role

of

Foundations

in

Documentation

Research

--

The Program

of

the

Council

on

Library

Resources",

Verner

W.

Clapp,

President,

Council

on

Library

Resources

Tuesday,

April

16

Coordinated

Systems

11.

Introduction:

"Foreign

vs.

American

Developments",

Allen

Kent,

Associate

Director;

Robert

E.

Booth, Research

Associate;

Center

for

Documentation and Communication

Research,

School

of

Library

Science,

Western Reserve

University

12.

"Intercontinental

Guided

Missives",

James

D.

Mack,

Librarian,

Lehigh

University

13. A

Corporation's

International

Network:

"Tentative

Proposal"

-

Subject

field,

"Petrole-

um"

-

Presented

by George S.

Crandall,

Philip

Q.

Stumpf,

Technical

Information

Group,

Research

and Development

Laboratory,

Socony Mobil

Oil

Company,

Inc.

14.

The

F.

B.

I.

Network -

Subject

field,

"Operations

ot

the

Identification

Division"

-

Presented by

C.

Lester

Trotter,

Assistant

Director,

Federal

Bureau

of

Investigation

15.

Library

Networks -

Subject

field,

"Any"

-

Presented

by

Herman

H.

Henkle,

Librarian,

John

Crerar

Library;

Margaret

E.

Egan, School

of

Li-

brary

Science,

Western Reserve

University

16. Communications:

"Present

and

Future"

-

Computers

and

Automation

Presented

by

R.

C.

Matlack,

Special

Systems

Engi-

neer,

Bell

Telephone

Laboratories,

Inc.

17.

Data-Vision:

"Video Communication

by

Telephone Line" -

Presented

by

J.

C.

Langner,

Electronics

Engineer,

Fitzgerald's

Communications

Systems,Using Accounting

or

Statistical

Machines

18.

Retrieval

of

Information

Manually

or

by

Machine -

Subject

field,

"Armor and

kinetic-energy

armor-defeating

alllmunition" -

Presented

by

John

McCafferty,

Chief,

Technical

Information

Section,

Watertown

Arsenal

19.

Control

of

Data

on

the

Pharmacologic

Properties

of

Chlorpromazine -

Subject

field,

"Pharmacology" -

Presented

by

Robert L. Hayne,

Fred Turim,

Science

Information

Department, Smith,

Kline,

& French

Laboratories

20. Chemical

Structures

and Responses

of

Or-

ganisms

to

Applied

Chemicals--Coordination

-

Subject

field,

"Biological

activity

and chemical

substances"

-

Presented

by

George

A.

Livingston,

Isaac

D.

Welt,

Chemical-Biological

Coordination

Center,

National

Research Council

21. Indexing and

Retrieval

of

Literature

Using

Machine-sorted Punched Cards -

Subject

field

-

"Fuel and

lubricant

additives"

-

Presented

by

Ben

H.

Weil, Manager,

Information

Services

Divi-

sion;

Barbara Hildenbrand,

Supervisor,

Literature

Searching

Section,

Information

Services

Division,

Ethyl

Corporation

Research

Laboratories

22. "The

Patent

Office

Problem", Robert

C.

Watson, Commissioner

of

Patents

Wednesday,

April

17

Systems Using Accounting and

Statistical

Machines

23. Machine

Searching

for

Legal Research

Subject

field,

"Mechanics

liens"

-

Presented

by

Frederick

B.

MacKinnon, John

C.

Leary, American

Bar Foundation

24.

Retrieval

of

Information

from

Technical

Reports

on

the

Development Problems

of

Various

Plastic

Products

-

Subject

field,

"Plastics"

-

Presented

by

Gilbert

L.

Peakes,

Bakelite

Company

25.

Adaptation

of

the

ASM-SLA

Metallurgical

Literature

Codes

for

Use

with

Machine-sorted

Punched Cards -Subj

ect

field,

"Metall

urgy" -

Presented

by

Barbara

H.

Weil,

E.

A.

Clapp,

Electro

Metallurgical

Company

26. A

Deep

Index

for

Internal

Technical

Re-

ports

-

Subject

field,

"Chemistry" -

Presented

by

Fred

R.

Whaley, Linde

Air

Products

Company

Systems Using Computers

or

Computer-Like Devices

27.

Abstracting,

Coding, and

Searching

the

Metallurgical

Literature

for

A.S.M.

The

WRU

-

32

-

Searching

Selector

-

Subject

field,

"Metallurgy"

-

Presented

by

Cedric

Flagg,

Research

Associate,

Allen

Kent,

Associate

Director,

Center

for

Docu-

mentation

and Communication Research, School

of

Library

Science,

Western Reserve

University

28. Machine

Searching

of

Patent

Files

using

the

SEAC

Computer (N.B.S.) -

Subject

field,

"Steroid

compounds" -

Presented

by

Don

D.

Andrews,

Director,

Research

and Development,

U.

S.

Patent

Office;

R.

A.

Kirsch,

Louis

C.

Ray,

National

Bu-

reau

of

Standards

29. Documentation

by

the

Filmorex Technique -

Subject

field,

"Scientific

information"

-

Presented

by

Jacques Samain, Chef de

Service,

Centre

National

de

la

Recherche

Scientifique,

Paris,

France

30. A Minicard System

for

an

Information

Center

-

Subject

field,

"Any" -

Presented

by

J.

W.

Kuipers,

A.

W.

Tyler,

W.

L.

Myers, Eastman

Kodak

Company

31. Panel

Discussion:

"Machine Li

terature

Searching

Potentials

in

a

Variety

of

Subject

Fields"

-Moderator, Robert

C.

McMaster,

Professor

of

Engineering,

Ohio

State

University

-

END-

*

----------

*

-----------*

BULK

SUBSCRIPTION RATES

These

rates

apply

to

prepaid

subscriptions

coming

in

together

direct

to

the

publisher.

For

example,

if

7

subscriptions

come

in

together,

the

saving

on

each

one-year

subscription

will

be

24

percent,

and

on

each

two-year

subscription

will

be

31

percent.

The

bulk

subscription

rates,

depending

on

the

num-

ber

of

simultaneous

subscriptions

received,

follow:

Number

of

Simultaneous

Subscriptions

7

or

more

4

to

6

3

2

Bulk

Subscription

Rates

(United

States)

Rate

for

Each

Subscription,

and

Resulting

Saving

to

Subscriber

One

Year

Two

Years

$4.20,

24%

4.60,

16

5.00,

9

5.25,

5

$7.25,

31%

8.00,

24

8.80,

16

9.55,

9

For

Canada,

add

50

cents

for

each

year;

outside

of

the

United

states

and

Canada,

add $1.00

for

each

year.

Forum

ON

GLOSSARIES -

AND

MALAPROPISMS

Alston

S.

Householder

Mathematics

Research

Center,

U.

S.

Army

Madison 6, Wisc.

A

common

symptom

of advancing age

is

a

growing

impatience

with the foibles,

fancies,

and

failings

of

the younger

generation,

coupled with a

release

of the inhibitions

that

had

earlier

restrained

possible

hortatory

impulses.

Age and

experience

may

not

always

bring

wisdom

to

others,

but

it

is

axiomatic

that

they

do to

oneself

I

Among

my

own

favorite

abominations, a

dis-

respect

or

low

regard

shown the

mother

tongue

ranks

very

high.

In

particular,

we, the

mathema-

ticians

and the

processors

of

data

and

informa-

tion,

are

presumably

experts

in

the handling of

symbols.

What,

for

example,

are

flow-charts

and coding

sheets?

Presumably

each

of us

is

meticulous

in

manipulating the

special

symbols

of

his

trade.

Why, then, should he be

less

so

in

handling

those

symbols

that

are

the

common

prop-

erty

of

the

public

at

large,

experts

included, and

especially

those

which

relate

more

directly

to

his

own

particular

domain?

Yet

examples

of

laxity

are

legion. One

can

argue,

of

course,

that

the

English

language,

like

the

Chinese people

through-

out

history,

combats

invaders

by aSSimilating

them.

But while the analogy

is

possibly

apt,

it

should be equally

apt

to

remark

that

assimilation

occurs

only when

other

defenses

have

failed.

On the

technical

level,

a

rather

recent

assim-

ilation

is

the

term

"algorithm",

defined

as

"any

peculiar

method

of computing".

In

just

commemo-

ration

of the

Arabian

algebraist

al-Khwarizmi,

the

term

"algorism"

was

adopted into the language

to

denote the

art

of computation with

Arabic

nu-

merals.

But

alas

for

the

memory

of the defunct

algebraist,

logarithms

also

have

to

do

with

com-

puting and the

terms

have a

certain

superficial

similarity.

Hence

the

mmning

was

broadened

and

the

word

transformed,

and

algorisms

survive

now

chiefly

in

the

dictionary.

Sic

transit

gloria

mundi.

The

bastard

formation

"eigenvalue"

is

rather

too

technical

for

most

general

dictionaries

(I

under-

stand

Veblen pounced upon

an

offender

with

"You

mean

proper

Werte?

"),

but

it

is

probably

perman-

-33 -

entlyentrenched.

Latin-Greek

and

Greek-Latin

hybrids

are

fairly

common

and

seem

somehow'

less

objectionable

since

these

"dead"

languages,

in

barbaric

times

like

these,

tend

to

coalesce

into

a

uniform

nimbus.

But however objectionable,

"eigenvalue"

seems

to

be

here

to

stay.

Dictionaries

do not

yet

recognize

"parallel-

opiped",

but

they

may

be

forced

to, eventually.

The

fourth

vowel plays a

very

incidental

role

in

"parallelogram

";

it

provides

a euphonic

link

for

the

essential

elements

"parallel"

and

"gram".

Yet somehow

this

particular

vowel

assumes

a

prominence

incommens

urate

with

its

function,

and

tends

to

displace

the

second

"e"

in

"parallel-

epiped",

as

though the

word

were

composed

of the

parts

"parallel"

and

"piped".

Actually the

solid

figure

is

a

parallel-epi-ped,

or,

if

you

like,

a

parallel

upon a

pedestal.

Bipeds

exist,

but

pipeds

do

not

(or

not yet).

In

these

examples

the

assimilation

is

in

vary-

ing

degrees

complete,

and

probably

nothing

can

be

done about

it.

In

some

others

there

seems

to

be

a

trend,

but

whether

the

process

is

reversible

is

yet

to

be

seen.

Among data-process:>rs, a

misconstruction

of

"data"

seems

especially

reprehensible

but

there

is

a

singular

trend

toward

treating

it

as

singular.

1 have

not

yet

seen

an

intrusion

of

"datas"

but

I

am

waiting, and,

in

fact,

the

correct

singular

"datum

If

seems

to have gone

out

of

existence.

Manuals

issued

by a

leading

manufacturer

(which

shall

here

be

nameless

I)

seem

to

be

consistent

in

taking

"data"

to

be

singular.

There

are

nouns

which

can

designate

either

a

class

or

the

members

of

that

class,

hence which

can

be

construed

as

singular

or

plural

according

to the intention of

the

speaker

(or

writer);

and

there

are

nouns which,

though

plural

in

form

are

singular

in

construction,

but

"data"

is

not

(yet) among

them.

Will

it

be-

come

so?

Mathematicians

and the

processors

of

data

should

be

able to count,

at

least,

and we should

Computers and Automat1on

not

be,

even

if

we often

are,

corrupted

by the

col-

loquial

"between you and

me

and the

gatepost".

"Between"

refers

specifically

to two, and

not

to

three

or

more.

A

can

be

between any two of B, C,

and

D,

but

it

is

among the

three

of

them.

Not

being

a

lawyer,

I cannot

argue

with

the

lawyers

about

their

beloved

"and/or",

but

in

either

scientific

or

popular

discourse

it

can

be explained

only

as

affectation

or

laziness.

The

disjunction

expressed

by

"or"

is

not

necessarily

exclusive.

Hence

"x

or

y

or

both"

is

unobjectionable.

It

i&

moreover,

inexpensive

in

terms

of

time,

space,

and

printer's

ink, and

even

the

"or

both"

is

gener-

ally

unnecessary

except

for

special

emphasis.

A

fact

is

a

fact

and no

argument,

period.

And

yet

repeatedly

one

hears

phrases

like

"true

facts",

"correct

facts",

or

assertions

that

people

argue

over

whether

some

fact

is

so.

If

it

is

a

fact,

it

cannot

fail

to

be

so.

One

can

argue

about

theories,

assertions,

speculations,

hypotheses,

conjectures,

but

never

about

facts.

One

can

argue

over

whether

an

assertion

states

a fact,

or

whether

that

which

is

asserted

as

a

fact

is

indeed a

fact.

And one

can

adduce

facts

to

support,

negate,

refute

or

confirm

an

argument.

But

about

facts

one

cannot

argue.

And

that'

s a

fact.

Even

scientists

are

sometimes

subject

to a

linguistic

confusion of

cause

and

effect

in

succumb-

ing to an

epidemic

aphasia

in

the

use

of

"imply".

Perhaps

"infer"

sounds

more

elegant,

but

whatever

the

reason

the

custom

of using

it

in

the

sense

of

"imply"

seems

to be mounting

in

scientific

as

well

as

in

popular

discourse.

It

is

true

that

diction-

aries

recognize

this

as

a

possible

meaning

of

"in-

fer",

but

the

habit

is

still

reprehensible

since

it

creates

a

potential

source

of confusion.

Fowler,

the

English

lexicographer,

was

especially

emphatic

about the

importance

of

segregating

meanings,

not

permitting

a

word

having a

particular,

definite,

function,

to

encroach

upon

another

whose function

is

distinct.

Clearly

"infer"

and

"imply"

are

such

words.

There

is

a

story

that

a

certain

psychologist

would

not

speak

to

his

daughter

for

a

week

because

three

times

in

her

thesis

she

used

"different

than".

If

he

were

alive

today and behaved uniformly, he

might

condemn

himseH

to

much

silence.

Accord-

ing

to

Webster,

x

may

be

greater

than

y

or

less

than

y,

and'in

either

event

it

is

other

than

y, and,

in

fact,

different

from

y.

Under

"different"

it

is

stated

that

whereas

the

word

is

normally

followed

by

"from",

there

are

good

literary

precedents

for

following

it

by

either

"to"

or

"than".

But none

is

-

34

-

cited

and

the

use

is

said

to be

considered

improper

by

many.

The

fact

of the

matter

is,

one

does

not

often

see

the bald, unadorned

statement

"x

is

dif-

ferent

than

y".

Rather,

"than"

is

generally

intro-

duced

in

an

ineffectual

attempt

to

escape

from

a

grammatical

trap,

as

in

"a

different

x

than

y".

The tendency

to

fall

into

this

trap

is

understand-

able.

English, unlike

French,

commonly

places

the

modifier

before

the

word

modified. Hence

one

starts

with

"a

different

x"

and

then

realizes

that

the

modifier

is

incomplete.

One

might

like

to

say

"a

different-from-y

x",

but

everyone

sees

this

would be going too

far.

The

reasonable

es-

cape

is,

as

usual,

very

simple.

One

has

merely

to

start

over

and

say

"an

x

that

is

different

from

y",

or

even

"an

x

different

from

y".

In

colloquial

speech,

of

course,

one cannot

erase

and

may

not

wish

to

start

over,

but

in

writing

there

is

no

excuse.

There

are

a

number

of

pitfalls

for

the

non-

native

in

expressions

of

degree,

and often the

natives

succumb,

whether

through

imitation

or

mere

carelessness.

We

seem

on

the

way

to

adopting

"less

or

equal

to",

or

"equal

or

less

than",

or

both, and I

can

see

no

real

argument

against

either,

except

that

it

offends the

purist's

(at

least

this

purist's)

ear,

like

the

omission

of

a

bar

from

a

well

known tune.

There

are

diffi-

culties

in

the

use

of

words

like

"enough",

"suffi-

cient",

"adequate".

In

these

days

of inflation,

linguistic

as

well

as.

monetary,

one

sees

at

times

an

amusing

misuse

of

"excellence"

as

though

it

could be

used

as

a

criterion.

In

fact,

the

word

means

"surpassing

requirement

or

expectation",

and hence

stands

for

a judgment one

can

make

only

ill

retrospect.

One

can

hardly

make

an

ad-

vance

requirement

that

one's

requirements

be

exceeded.

At a

more

technical

level,

though, and

still

on

the

subject

of

relations,

I have

seen

many

people who should know

better

apply the

term

"equation"

where

they

should have

used

"expres-

sion".

And

even

more

often

"roots"

is

used

where

zeros

are

meant.

The

distinction

may

seem

pe-

dantic,

but

functions and polynomials

possess

zeros;

only

equations

possess

roots.

To

return

briefly

to

grammatical

traps,

a

possibly

harmless

but

rather

amusing

and common

one

starts

out with ''What

it

is,

is

•..

"

After

arriving

at

the

first

"is"

one

recognizes

the

need

for

the

second,

yet

feels

uncomfortable about the

repetition,

and

o~e

escape

is

to

disregard

grammar

and

drop

it.

The

logical

solution

is

again

to

start

over

with

a

simple

"it"

in

place

of ''what

it

is".

(Cont'd

on

ON

GLOSSARIES

(Cont'd

from

page 34)

Or

if

one

feels

that

more

of

an

introduction

is

needed,

it

would be: ''Well,

I'll

tell

you

what

it

is.

It'S

•••

"

A

true

redundancy

cannot

be

condemned

on

either

logical

grounds,

or

grammatical.

Hence

everyone

has

a

right

to

his

redundancy,

and

some-

times

it

can

be

effective,

as

for

emphasis.

But

a

particular

redundancy,

often

repeated,

comes

to

lose

its

force.

An

example

of

such

is

"each

and

every".

I

have

no

idea

who

started

the

fad,

but

in

my

mind

it

is

strongly

associated

with

in-

spirational

talks

heard

during

my

youth

in

YMCA's

and Sunday Schools.

In

technical

discourse

today

it

has

nothing

to

contribute.

But

"each"

and

"every"

suggest

"all", and

also

a

little

suggestion

I

heard

made

by

a

former

professor.

The

suggestion

is

that

it

is

generally

easier

to

speak

in

the

singular

than

in

the

plural.

If

a

property

P

is

possessed

by

all

members

of

a

class

C,

then

it

is

possessed

by

each

(and

every!)

member

of

C.

Many

times

I

have

started

to

say

something

about

all

members

of a

class,

and

be-

come

hopelessly

lost

in

a

grammatical

maze.

But

by

starting

over

and

making

the

assertion

about

each

(Q!

every)

member

the

difficulties

melted

away.

I

have,

in

my

own

published

writings,

perpe-

trated

and

propagated

many

barbarities

for

which

I

blush

today.

In

a

rapidly

changing

technical

field

the

language

is

correspondingly

fluid, and

all

of

us

who

publish

on

the

subject

are

helping

to

create

and

establish

the

technical

vocabulary

of

the

field.

We

have

therefore

a

particular

respon-

sibility

to

select

our

technical

vocabulary

well.

Among

the

words

I

have

helped

perpetuate

and now

regret,

are

"round-off"

and

"approximation"

used

as

adjectives.

The

correct

adjectives

are,

of

course,

"rounding"

in

the

one

case,

"approximate"

or

"approximating"

in

the

other.

It

is

part

of

the

strength

of

our

language

that

it

is

possible

to

use

a noun

as

a

modifier

when

needed,

although

there

are

times

when

one

could

wish

for

a

special

ending

to

be

attached

as

a

signal.

But

in

any

event,

to

use

a noun

as

an

adjective

where

a

perfectly

good

adjective

exists

is

surely

gratuitous.

And

so

my

homily

concludes

with

a

brief

con-

fession.

I

seem

to

be

fallible,

too.

-

END-

WESTERN

COMPUfER

CONFERENCE

*

(cont'd

from page

29)

*

conditions.

A

special

purpose computer,

built

for

the

Rome

Air

Development

Center

by

the

Monroe

Calculating

Machine

Company,

included

a unique

automatic

internal

~iagnosis

unit

which

quickly

pinpoints

any

malfunction

in

one

of

the

computers

4,500

logical

elements.

This

paper

will

discuss

the

diagnosis

tech-

niques

employed and

will

present

the

field

performance

results

achieved

in

the

pioneer

application.

SESSION

X:

PROGRAMMING

Paul Armer, The

RAND

Corp.,

Chairman

37. "The

Variable

Word

and Record Length

Problem and

the

Combined Record Approach

on

Electronic

Data

Processing

Systems",

N.

J.

Qean, Ramo-Wooldridge Corp.

--

This

paper

distinguishes

between

"fixed",

"adjustable"

and

"variable"

word and

record

lengths.

It

discusses

some

of

the

advantages

of

the

vari-

able

word and

record

lengths

and

the

"expand-

able"

record

in

practical

business

data-pro-

cessing

applications.

A

technique

for

reduc-

ing

the

variability

of

storage

required

for

detailed

transactions

is

presented.

A

typi-

cal

application

(commercial

deposit

account-

ing)

is

described

in

which a

drastic

reduction

in

storage

requirement

is

affected

by

utiliz-

ing

st~tistical

averaging.

38.

"Empirical

Exploration

of

the

Logic

Theory Machine",

A.

Newell,

J.

C.

Shaw,

H.

A.

Simon,

The

RAND

Corp.

--

The Logic Theory

Machine

is

a program

that

discovers

proofs

for

theorems

in

elementary

symbolic

logic.

It

does

this,

not

by

means

of

an

algorithm

(although

such

algorithms

exist),

but

by

using

heuristic

devices,

much

as

a

human

does.

This

paper

presents

the

results

of

detailed

explor-

ations

of

the

program

on

RAND'S

JOHNNIAC

(see

following

paper).

It

describes

the

program

and

evaluates

the

contribution

of

the

various

methods and

heuristics

to

the

total

problem

solving

capability

of

the

machine.

39. "Programmi'ng

the

Logic Theory Machine",

A.

Newell and

J.

C.

Shaw, The

RAND

Corp.

--

The

Logic Theory Machine

(called

LT,

see

represents

a

class

of

non-numerical

problems

with

quite

different

programming

re-

quirements

than

either

normal

arithmetic

cal-

culation

or

business

data

processing.

The

program

itself

is

a

large,

complicated

hier-

archy

of

subroutines.

For

LT

an

intermediate

language

(interpretive

psuedo code)

was

written

for

the

RAND

JOHNNIAC.

This

language

is

inde-

pendent

of

symbolic

logic,

the

subject

matter

of

LT,

and

is

a

general

language

for

informa-

tion

processing.

This

paper

first

character-

izes

the

programming problems

involved,

and

then

illustrates

solutions

to

them

by

describ-

ing

the

language.

-

END

-

35

-

DIVISION

OF

LABOR

(cont'd

from

page

7)

Computers and

Automation

The

details

of the

work

beyond the pro-

grammer

will

not

be

discussed

in

this

article.

The

people

r~quired

to

handle

these

details

are

included

'in

the bottom block of the

ac-

companying diagram.

Some

of

these

details

are:

data

transition

from problem

to

form

acceptable

for

machine

input,

card punching,

routine

running of the problem

on

the machine

for

various

data

input

sets,

and the

clerical

work involved in any

operation.

What

has been

described

(as diagrammed

in

the accompanying

chart)

'should complete

the

requirements

for

a complete computer

in-

stallation.

The

number

of people

will

depend

on

the

size

of the

installation.

If

it

is

so

desired,

a

unit

of the type

shown

on

the

dia-

gram

may

specialize

in a

particular

applica-

tion,

such as

aircraft;

then the complete

or-

ganization

may

be

made

up

of a

unit

for

each

application

in

which the

company

is

engaged.

,---

t

~,-_S_U_P_E_R_V_IS_O_R_-,~

- - - - - - - -

-I

t

~

--

----...

MACHINE

COMPUTER COMPUTER

---

TECHNIQUES

APPLICATION

SPECIALIsT

CONSULTANT

I 1 L I

+

...

NUMERICAL

NUMERICAL NUMERICAL -

ANALYST ANALYST ANALYST ANALYST

PROGMMR.

~

....

PROGMMR.

I-

.....

PROGMMR.

~

r-

PROGMMR.

~

PROGMMR.

~

L..e

PROGMMR.

l-

I-

PROGMMR.

~

F-

PROGMMR.

~

CLERICAL,

DATA

TRANSCRIPTION,

KEY PUNCH, MACHINE OPERATION

AND

PRODUCTION

ORGANIZATION

for

CHART

COMPUTER SERVICE SECTION

-36 -

COMPUTER

USE

Ccont'd from page 19)

and examined

by

a simple program which

pro-

duces

the

following

effect:

(a)

Read

in

the

next symbol.

(b,l)

If

it

is

an

alphabetic

character,

and

if

the previous

punctuation

mark

has been doubled, punch

it

out.

(b,2)

If

it

is

an

alphabetic

character,

but

the

mark

has not been doubled,

return

to

(a).

(b,3)

If

it

is

a

punctuation

symbol

examine

the

If

this

is

also

a

punctuation

symbol

store

suitable

instructions

to

produce

(b,l)

and

(b,2).

Punch

out the

symbol

and

return

to

(a).

If

the next

character

is

not a

punctuation

symbol

modify

(b,l)

and (b,2)

appropriately

and

return

to

(a).

The

final

operation

is

to

print

out

the

contents

of

the

tape

just

produced, again

presented

to

the

printing

reader

backwards.

The

output

will

be a

list

of

the

iso-structural

units

contained

in

the

original

text.

This type

of

program can

be

extended

to

give

page and

line

indications,

and

also

to

enable simultaneous examinations

to

be

made

for

several

different

structural

units.

Since,

however,

no

ne~

principles

are

involved,

it

is

not worth

detailing

the

program

steps

which

are

required.

BIBLIOGRAPHY

(1) "Nature",

by

A.

D.

Booth, 176, (1955) 565.

(2) "Automatic

Digital

Calculators",

by

A.

D.

Booth and

K.

H.

V.

Booth,

p.

221,

2nd

Ed.

Butterworths,

London

(1956).

*------------------*-------------------*

NEW

PRODUCTS

AND

IDEAS

(~ont'd

from page

9)

he

states

his problem

as

an equation. Next, he

draws

up

a

flow

chart

analysis of the problem as

a

series

of computer operations. Then he fills out

a coding sheet from which the computer's key punch

operators

take instructions.

Finally, he

prepares

a

test

case

to which he

knows the answer.

When

his problem runs

on

the

machine, he checks the answer, and eliminates

troubles through diagnostic methods taught in the

class.

The Lockheed do-it-yourself computing oppor-

tunity

is

the

result

of

a

new

simplified coding

tech-

nique called

SOAP

(for Symbolic Optimum Assembly

Programming) developed by International Business

Machines Corp. Using this technique, the employee

can write the actual equations

on

his coding sheet

instead

of

a more complicated system of address

numbers normally used to locate various pieces of

information within the machine. The

new

oppor-

tunity

is

designed for employees

who

have never

seen

or

used a computer before.

It

can be used

to solve both scientific and accounting problems.

*

-37 -

Are

you

looking

for

IDEAS

AND

KNOWLEDGE

about the Computer

field?

Here

are

some

of the

articles

and papers which

we

have published

in

recent

issues

of "Computers and

Au

toma

tion":

November:

Use

of

Automatic Programming

--

Walter

F.

Bauer

Data Problems of a Grocery Chain

--

Frank

A.

Calhoun

The

Power

of the Computer

--

George

J.

Huebner,

Jr.

An

Automatic Micro-Image

File

--

National

Bureau of Standards

December: Indexing

for

Rapid

Random

Access

Memory

Systems

--

Arnold

I.

Dumey

Self-Repairing

and Reproducing Automata

--

Richard L. Meier

The

Computer's Challenge

to

Education

--

Clarence

B.

Hilberry

January.

1957

(vol.

6, no.

1):

Modern

Large-Scale

Computer System Design

--

Walter

F.

Bauer

Logical

and

Combinatorial Problems in Computer

Design

--

Robert

McNaughton

Transistorized

Magnetic Core

Memory

--

Bell

Telephone

Laboratories

Education

for

Automation

--

Alston S. House-

holder

Social

and

Public

Relations'

Responsibilities

of

the Computer

Industry

--

Jay

W.

Forrester

High School Science Education

--R.W.

Melville

February: Computation

for

an

Earth

Satellite

--

Neil

D.

MacDonald

New

Computer Developments

Around

the World

--

Everett

S.

Calhoun

Industry

and

the

Automated

Future:

Problems Along

the

Way

--

John Diebold

Electronic

Digital

Data-Handling

--

Howard

T.

Engstrom

The

Solution

of Boundary Value Problems

on

a

REAC

Analog Computer

--

M.

Yanowitch

March:

Office

Equipment Outlook

--

Oliver

J.

Gingold

New

Products and Ideas

Group Behavior

of

Robots

--

Manfred

Kochen

Robots

and

Automata: A

Short

History

--

James

T. Culbertson

Automatic Computing Machinery

--

List

of Types

Components

of

Automatic Computing Machinery

--

Lis t

of

Types

The

Computer

Field:

Product~

and

Services

for

Sale

or

Rent

List

of Headings

BACK

COpy

PRICE:

If

available,

$1.25 each,

except

June 1955,

$4.00,

and June 1956, $6.00 (the June

issue

if

the

Computer

Archives

of

Neurology

and

Psychiatry,

Vol. 67,

Feb.

1952

232.

Brazier,

Mary

A. B. ;

Expanding

Concepts

in

Neurophysiology,

Archives

of

Neurology

and

Psychiatry,

Vol. 67,

April,

1952

233.

Gerard,

R.

W.

; The

Biological

Roots

of

Psychiatry,

Amer.

Jr.

of

Psychiatry,

Vol. 112,

No.2,

Aug. 1955

-

END-

PHYSICISTS

aDd

MATHEMATICIANS

How

do

your

special

skills

fit

into

the

Aircraft

Nuclear

Propulsion

picture

at

General

Electric

Many physicists

and

mathematicians who recognize

the

exceptional promise

of

a

career

in

General

Electric's

fast growing Aircraft Nuclear Propulsion Department

have asked

if

their skills

can

be

used

in this significant

project.

11

you

are

qualified

to

work

on:

Thermodynamic

and

Air

Cycle

Analysis

Reactor

Analysis

Shield

PhYSics

Nuclear

Instrumentation

Applied

Mathematics

Digital

and

Analog

Computer

Theoretical

Physics

you

can

move now into major assignments

in

the

de·

velopment

of

nuclear

propulsion systems for aircraft.

You do not

need

experience.

Through

General

Electric's full tuition

refund

plan

for

advanced

university courses

and

inplant training

conducted

by

experts, you'll

acquire

the

necessary

nucleonics knowl·

edge.

The field itself assures you a

rewarding

future, but, more

than

that, the physicist

or

mathematician who likes to

work

in

a top·level scientific atmosphere will

appre·

ciate

General

Electric's

encouragement

of

creative

thinking, its recognition of accomplishment.

Comprehensive benefit program • Periodic merit reviews

Excellent starting salaries • Relocation expenses paid

Openings

in

Cincinnati,

Ohio

and

Idaho

Falls,

Idaho

Write

in

confidence, stating salary requirements,

to location you prefer:

Mr.

J.

R.

ROSSELOT

P.

O.

Box 132

Cincinnati,

Ohio

I

Mr.

L.

A.

MUNTHER

I

P.

O.

Box 535

I

Idaho

Falls,

Idaho

I

GEN

ERAL

e

ELECTRIC

*

-39

PLEASE

RUSH

your

specially

J)repared

1957

GE

Tantalytic ® Bulletin listing

over

250

Tantalum

Capacitors

in

numerlca!

sequence

by

microfarads.

COMPANY

NAME

••••••••••••••••

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

ATTENTION

•••••••••••••••••••••

Are

you

Looking

for

REFERENCE

INFORMATION

in the

Computer

Field?

COMPUTERS

AND

AUTOMATION

publishes

16

kinds of reference, information

that

computer people can hardly

afford

to

be

without.

Latest

issues

containing each

are

indicated.

Organizations:

Roster

of

Organizations in the

Computer

Field

(June,

Aug.

1956)

Roster

of

Computing Services (June

1956)

Roster of Consulting Services (June

1956)

Computing Machinery

and

Automation:

Types

of

Automatic

Computing

Machinery

{March

1957>

Roster

of

Automatic Computers (June

1956)

Outstanding

Examples

of Automation {July

1954

Commercial Automatic Computers

(Dec.

1956)

Types

of

Components

of Automatic

Computing

Machinery

{March

1957>

Products

and

Services in the Computer

Field:

Products

and

Services

for

Sale

or

Rent (June

1956)

Classes

of

Products

and

Services

{March

1957>

Words

and

Terms: Glossary of

Terms

and

Expressions

in

the

Computer

Field

(Oct.

1956)

n.

Information

and

Publications:

Books

and

Other

Publications

(many

issues)

New

Patents

(nearly every issue)

Roster

of

Magazines

(Dec.

1955)

Titles

and

Abstracts of Papers Given

at

Meet-

ings

<many

issues)

People:

Who's

Who

in

the

Computer

Field

Cumulative (1957, Extra

No.

ED

Supplements (various issues)

BACK

COPY

PRICE:

If

available,

$1.25 each,

ex-

cept:

June 1955, $4.00,

and

June 1956, $6.00

(the June

issue

is

the

Computer

195704

195704 195704

Navigation menu

Versions of this User Manual:

Views

Navigation