195705

195705 195705

User Manual: 195705

Open the PDF directly: View PDF .
Page Count: 48

Download
Open PDF In Browser	View PDF

Computer Market Survey -

Report No. 1
. . . Edmund C. Berkeley

New Products and Ideas

Reliability in Business Automatic Data Processing
Herbert T. Glantz
Automatic Search of Library Documents
. S. Richard Moyer

I. R. E. National Convention, March, 1957, New York Titles and Abstracts of Papers Bearing on Computers
and Data Processors

As recently as ten years ago it was just becoming
evident that digital techniques in electronics were
destined to create a new and rapidly growing field.
Today, incorporat~d in electronic compu,ters and
other equipment, they constitute one of the most significant developments in' scientific computation, in
electronic data processing for business and industry,
and in electronic cantlol s~stems for the military. In
the near future they are expected to become a major
new factor in industrial process control systems.
The digital computer for scientific computation is
becoming commonplace in research and development
laboratories. Such machines range from small specialized units costing a few thousand dollars, to large
general purpose computers costing over a million
dollars. One of these large computers is a part of the
Ramo-Wooldridge Computing Center, and a second
such unit will be installed the latter part of this year.
The digital computer has not only lightened the computation load for scientists and engineers, but has
made possible many calculations which previously
were impracticable. Such computers have played a
major role in the modern systems engineering
approach to complex problems.
Electronic data processing for business and industry
is now well under way, based on earlier developments
in electronic computers. Data processors have much

in common with computers, including the utilization
of digital techniques. In this field, teams of RamoWooldridge specialists are providing consulting services to a variety of clients on the application of data
processing equipment to their problems.
The use of digital techniques in military control
systems is an accomplished fact. Modern interceptor
aircraft, for example, use digital fire control systems.
A number of Ramo-Wooldridge scientists and engineers have pioneered in this field, and the photograph
above shows a part of an R-W-developed airborne
digital computer.
These, then, are some of the aspects of the rapid
growth which is taking place in the field of digital
techniques. Scientists and engineers with experience
in this field are invited to explore openings at The
Ramo-Wooldridge Corporation in:
Automation and Data Processing
Digital Computers and Control Systems
Airborne Electronic and Control Systems
Guided Missile Research and Development
Electronic Instrumentation and Test Equipment
Communication Systems

The Ramo-Wooldridge Corporation
5730 ARBOR VITAE STREET

•

LOS ANGELES 45, CALIFORNIA

COMPUTERS
CYBERNETICS
Vol. 6,

No.

•

AUTOMATION

AND

ROBOTS

•

AUTOMATIC

CONTROL

May, 1957
ESTABLISHED SEPTEMBER, 1951

ARTICLES
Computer Market Survey - Report No.1
Reliability in Business Automatic Data Processing
Automatic Search of Library Documents

Edmund C. Berkeley
Herbert T. Glantz
S. Richard Moyer

8
20
24

NEW PRODUCTS AND IDEAS
REFERENCE INFORMATION
1. R. E. National Convention, March 1957, New York - Titles and
Abstracts of Papers Bearing on Computers and Data Processing
New Patents
FORUM
Magazines on "Computology"
The United States Mails - Field for Automatic
Processing of Information
Analogue Computers in Europe
Association for Computing Machinery Meeting

... Samuel J. M. England
... The Reader's Digest and
the U. S. Post Office
P. A. R. Wright
... J. F. Summers

30
36

37
38
42
44
6
6
46

The Editor's Notes
Index of Notice s
Advertising Index

Edi tor: Edmund C. Berkp.ley
Assistant Edi tors: Neil D. lJacdonald, F. L. Walker
Contributing Editors:
Andrew D Booth, John W. Carr,III, Alston S. Hous~holder
Advisory Committee: Samuel B. Williams, Herbert F. Mitchell Jr., Howard T. Engstrom,
Alston S. Householder, H. Jefferson Mills, Jr.
Sales and Service Manager:
Miltnl'> L. Kaye, 601 Madison Ave., New York 22, N.Y. - Plaza 5-4680
Advertising Representatives:
New York - Mil,ton L. Kaye, 601 Madison Ave., New York 22, N.Y., Plaza 5-4680
San Francisco - W.A. Babcock, 605 Market St., San Francisco 5, Calif., Yukon 2-·3954
Los Angeles - Wentworth F. Green, 439 So. Western Ave., Los Angeles 5, Calif., Dunkirk 7-8135
Elsewhere - the Publisher
Publisher:
Berkeley Enterprises, Inc., 815 Washington St., Newtonville 60, Mass., Decatur 2-5453 or 2~3928
COMPUTERS ANn AUTOMATION is published monthly. Copyright, 1957, by Berkeley Enterprises, Inc.
'Subscription rates: in the United States - one year $5.50, two years $10.50; in Canada - one year $6.00, two years $11.50;
elsewhere - one year $6.50, two years $12.50
Entered as second class matter at the Post Office, New York, N.Y.

- 3 -

New Sylvania package offers

Maximum Dissipation
in Miniature
Diodes
Cooler operation resulting from
higher qissipation of Sylvania glassto-metal miniature diode permits
closer printed board spacing for
maximum savings in space.

Right angle bending of leads for
printed board insertion does not
affect the diode body since metal-toglass design avoids chipping or
cracking.

v
V
./

l/
V'
.. V
./v

Actual comparison of Sylvania miniature diodes
with all-glass miniatures shows that the Sylvania
metal-to-glass package design results in greater dissipation. As a result, cooler operation can extend
diode life aiid improve product dependability and
performance. Diodes can be banked closer on printed
circuit boards for maxim~m space savings.
Metal-to-glass package offers other important advantages. The diode cartridge is assembled before
installation of the whisker and d~e-avoiding excessive heating. In addition, right angle bending of
the leads for printed board insertion does not result
in chipping or cracking of the diode body.
Sylvania offers a complete line of miniature diodes
in the glass-to-metal package'. The package meets
the standard RETMA outline of .105" maximum
diameter and .265" maximum overall length. Write
for complete details .

.
20

JUNCTION TEMPERATURE RISE, TOC,

Typical dissipation curve of the
Sylvania glass-to-metal diode.

SYLVANIA ELECTRIC PR(>DUCTS INC.

1740 Broadway, New York 19, N. Y.
In Canada: Sylvania Electric (Canada) Ltd.
Shell Tower Bldg., Montreal

LIGHTING

RADIO

TELEVISION

ELECTRONICS

ATOMIC ENERGY

;1,

l
//

I'
J

ELECTRICAL CHARACTERISTICS OF SYLVANIA MINIATURE DIODES AT 25 0 C

Maximum
Forward Voltage

Minimum
Reverse
Resistance

Maximum
Reverse Recovery
@ 0.3 usee
(Note 3)

Stability

IN631

3.5 v
(Note 4)

500 kohms
(Note 5)

500 ua

Note 7

IN632

1V
If = 7 ma

500 kohms
(NoteS)

800 ua

Note 7

IN633

lV
If = 125 ma

500 kohms
(Note 6)

1650 ua

Note 7

Type

MinimU'm
Forward Current
at 1 volt

Maximum
Reverse Current

Minimum
Peak Inverse Voltage
(0 dynamic impedal;1ce)

IN67A

4ma

50 ua @ -50 volts
5 ua @-5 volts

100 volts

IN90

5 ma

750 Oa @ -50 volts

75 volts

;,)N98

20ma

100 ua @ -50 v
8 ua @-S v

IN126

Sma

850 ua @ -50 v
SO ua @ -10 v

75 volts

IN127

3ma

300 ua @ -50 v
25 ua @ -10 v

125 volts

10 ua @ -10 v

IN128

3ma

IN191

Sma

IN198

4ma
(5 ma @ 75° C)

IN634
IN635

Note 1
50 ua @ -50 v
(Note 2)
10 ua @ -10 v

100v

50 volts
Note 1
100 volts

SOma

45 ua @ -45 v
100 ua @ -100 v

115 volts

SOma

175 ua @ -150 v

165 volts

Note 1: For type 1 N191 at 55° C the
reverse resistance will be 400 ohms or
greater between -10 and -50 volts
when swept from 0 to -70 volts at a
60 cycle rate.
. The reverse recovery time will not ex·
ceed 0.5 usec at 700 ua or 3.5 usec at
87.5 ua of current when rapidly switched
, (at a 60 cycle rate) from +30 ma for-

ward current to -35 volts.
Nate 2: For type IN198 at 75° C the
maximum reverse current at -50 volts is
250 ua and at -10 volts is 75 ua.

Note 3: a) Forward current exposure
5 mao b) Reverse test voltage = 40 ±2
volts. c) DC circuit resistance = 2000
ohms.

Note 4: Peak measurement with half
sine wave of 50 ma peak current, 0.1
usec pulse width, and 100 kc pulse rep·
etition frequency.
Note 5: Minimum resistance in thou·
sands of oh'ms when Ell characteristic
is swept at 60 cycles from 0 to -70 volts
and resistance slope observed between
-10 and -60 volts.

Note 6: Minimum resistance in thou·
sands of ohms when Ell cha'racteristic
is swept at 60 cycles from 0 to -100
volts and resistance slope observed be·
tween -20 and -90 volts.
Note 7: Additional control measurements are made for reverse current hys·
teresis, reverse current drift, and flutter.

"Sylvania-synonymous with ~Semiconductors"

THE

EDITOR'S

BARRIERS TO COMMUNICATION
I.

The Computation Laboratory of Harvard University held "An International Symposium on the
Theory of Switching", April 2 to 5, in Cambridge,
Mass. As the conference began, Dr. Howard H.
Aiken, Director of the Computation Laboratory,
read some cables. The first one was from Dr.
Michael A. Gavrilov, 'Professor, Doctor of Technical Sciences, Academy of Sciences, Moscow,
U. S. S. R., saying that illness prevented him from
coming; his paper was "Investigation of Switching
Theory in the Soviet Union". The second cable
was from the Institute of Mathematical Machines,
Academy of Sciences, Prague, Czechoslovakia,
saying "Svoboda can not come"; Dr. Antonin
Svoboda, Director of the Institute, was to give a
paper, "Some Applications of Contact Grids".
Dr. Aiken said that the other speakers from the
u. S. S. R. (G. N. Povarov, Academy of Sciences,
Moscow, whose paper was" A Mathematical Theory for the Design of Contact Networks"; Alexander G. Lunts, Associate Professor of Mathematics, Leningrad Electrotechnical Institute, Leningrad, whose paper was "Network Equivalent Transformations by Means of Characteristic Functions";
and V. N. Roginskii, Senior Research Associate,
Academy of Sciences, Moscow, whose paper was
"A Graphic Method for the Synthesis of Contact
Multipoles") had not been heard from, and had
not arrived at the conference. He expressed the
regret - his own and of all those present - that
none of these welcome scientists from the U. S. S.
R. and Czechoslovakia had been able to come; and
he hoped that Dr. Gavrilov's illness was not contagious. Dr. Aiken added that the texts of all papers had been received, would be read at the scheduled place in the program, and would be published;
but that it was a great pity that discussion of questions with the author s could not take place.
We think we speak for all computer people
when we say we also deeply regret political considerations spreading contagiously into scientific
fields. We cannot see any earthly use to these
stupid barriers to scientific communication. We
are all of us human beings jointly pursuing scientific search to find out the secrets of nature.

II.

NOTES

Symposium, and looked at a great many new faces,
how much I might have in common with many of
them, if only I had a short cut for the first five minutes of conversation - when one decides whether
or not "this man is intere sting to me; we should
talk together".
So far, the be st form of communication at
any computer meeting seems to be the cocktail
party: no one sits down; everyone is mobile; you
can wander around, say many hellos, introduce
many people, and be introduced to many.
But there ought to be much better techniques
of communication at computer meetings: (1) The
tag which you wear at a meeting should show: your
name; your organization; your main interests. This
would give a wonderful starting point. (2) In your
pocket you should have a thumbnail sketch of yourself, a "Who's Who" report, - so that while you
read your new acquaintance's, he can read yours.
If computer people suffer from lack of communication, we have mainly ourselves to blame:
we are supposed to be experts in information handling, information engineer s I

*---------- * ----------*
NEW

PRO,DUCTS

AND

IDEAS

A Three-Axis Simulator for Controlled
Flight Test of Airborne Systems
2. A New Radar Device for Corrected
Mapping
3. Lattice Networks for Digital Storage
4. A New Idea in Component Supply
1.

*
INDEX

For Information on:
Advertising Index
Advertising Rates and
Specifications
Back Copies
Bulk Subscription Rates
Manuscripts
Reader's Inquiry Form
Special Issues

Page
18
19
19
19

*
NOTICES
See Page:
46
see Apr. issue
see Apr. issue
see Apr. issue
see Mar. issue
46
see Mar. 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 allow thr~e
weeks for the change.

But this kind of barrier is not the only kind
of a barrier. I thought as I walked around the
- 6 -

........
....................
••••••••
••••••••
••••••••
•
........
• ••• •
••••••••
••••••••
.........
••••••••
....
...
'
.
•••••
••
••• •• •
••••••••
••••••••
••••••••
••••••••
.••••••••
.......
• ••••
·••••••••
... '.
••••••••
••••••••
•••••
••
••••••••
..
..
••••••••
...
..
,.'·...
....
,..
..
.
....
..
••••••••
,

e'.'••• '.'." •
'

e, ,••• , •

••••

•• •

...
·......
...
.
......
.
•
.......
.
.'.......
....
.
.,••••••••
......
.
••••••••
.'...'.....
...
'.
.
.'
..
,.'
........'..
.,.'....
......
'.....
..
.
·......•.
.
..
.
..........
............
... ..
...
....." ...........
..·.....
'., ......
••••••••
.....
••••••••
...••••••••
..••••••••
......
·.......
'.", ..•.
.
....
,.'
......".,' '..,
.......
'.,
·.......
.'
...
••••••••
....
...
••••••••
,. '." .. ,.,
"................
~'

••••••••
• ,. •• ',a
,

NEW
HI-SPEEi>
SWITCHING TRANSISTORS
Assures Computer Reliability
Computer engineers long seeking NPN transistors in applications requiring high current and fast switching will specify
General Transistor's new 2N356, 2N357, and 2N358 for peak
reliability.
2N358: As developed by General, a typical switching speed of
.4 of a microsecond at 300 milliamperes of collector current
is possible with only 15 rna. of drive current.
The series resistance of these GT transistors. when conducting,
is V2 ohm; the nonconducting series resistance is as high
as 5 megohms with a result that approaches optimum efficiency
at high current levels .
Computer manufacturers know they can depend on General's
engineering and development as well as their quality and
service. That's why GT is the largest supplier of transistors
for computers.

•••••

•••••••

2N358 CHARACTERISTICS
Parameter
Conditions
Min
Typical Max
Collector-Base
Ic = 25Ila_ _ 20V__30V
Voltage
Vcb = SV
51la
Collector Cutoff
Ic = 300ma
Current
_ Vce=.2SV___ 20 _ _ 30 _ _ S0
D.C. Current Gain _ _ lc= Ima
gmc
Alpha Cutoff
Vcb = SV
Frequency_ _ _ lc= Ima
gmc

Ie a300Ma.

lSI ·I~

Ma.

out In" 5 Ma .

'~.,

• • '<, . '• • •'
',

;

'",

••••••••
~

:~~::: ~:,

Write for GT's special NPN Computer Transistor
Specifications Bulletin.

GENERAL TRANSISTOR CORP•
Jamaica 3S, N. Y.-OLympia 7-9700
Cable: Transistor New York

COMPUTER

MARKET

SURVEY -

REPORT

NO.1

Edmund C. Berkeley
Edi tor, "Computers and Automation"

INTRODUCTION

was requested "for statistical purposes" only.
(2) Type of Organization.
explanatory.

At the beginning of March, we mailed out
to our subscribers in the United States and
Canada a survey form, the main purpose of
which was to make an estimate of the market
for computer products and services. A copy
of the survey form appears on page 40.
Approximately 2200 of these forms were
mailed out; and as of the middle of April,
about 200 forms, or 9 percent, had been returned. The replies contain interesting, important, and extraordinary information, showing the nature and size of the current market
for computing and data processing machinery.
The replies will furnish the material for
several reports to our readers, particularly
after they have been carefully projected,
industry by industry, to determine their probable implications.
In the meantime, we think that the interests of our readers will be best served if we
publish a serial tabulation of a part of the
raw data received for each reply. This is
especially true because often in surveys the
examples of the raw data are just as interesting as summaries of the data. The tabulation
given below includes all replies except a few
incomplete or inapplicable returns.
Each entry in the tabulation is important,
because it represents a report by a subscriber
of "Computers and Automation" (whose title in
his organization has been indicated) who kindly
took the trouble to serld us in his estimate for
the advantage of others. We express our app~
ciation to these subscribers of ours for sending us this information.
WHAT THE TABULATION CONTAINS
An explanation of the information in the
table follows:
'
(1) Code. This code identifies the return, not the organization. Although the name
of the organization was given on nearly every
return, we cannot publish the name since it
- 8 -

This is self-

(3) Size. Size is reported in number of
employees, grouped. S stands for small size,
I to 50 employees; M, medium size, 51 to 500
employees; L, large size, 501 to 5,000 employees; vL, very large size, over 5,000 employees.
(4) Recom., Title. Here is the report on
the question "Do your recommendations affect
purchases?", and the title of the person filling in the questionnaire. The purpose of this
information is to give some indication of the
probable authority of the estimate here reported.
(5) Products and Services. Here is the
report on the question "What kinds of computer
products and services does your organization
buy or rent (or is considering buying or renting?"
Column (a) reports on "automatic digital
computers".
Column (b) reports on "automatic analog
computers".
Column (c) reports on "other computers";
the codes are as follows:
S - simulators
.
v - other data processing machines
Column (d) reports on "components"; the
codes are as follows:
D - Delay lines
M - Magnetic tape devices
T - Transistors
o ~ Other components
Colum~(e) reports on "services"; the
codes are as follows:
P - comPuting services
S - conSulting services
o - Other services
(6) Likely Spending Next Five Years.
This column reports on the question:
"Can you estimate roughly and approximately
about how much your organization is likely to
spend on products and services in the computer field in the next five years?
Between
and_____ "

Computer Market

If a lower figure was given but no upper figure,
'we have inserted for the upper figure an amount
equal to twice the low~r figure. If this question was not answered, but the "likely spending" in the next twelve months was estimated,
we have taken the one year figures and multiplied by five.
A ROUGH AND APPROXIMATE ESTIMATE OF THE MARKET
FOR COMPUTER PRODUCTS AND SERVICES

wind up with a rough and approximate overall
estimate of the computer market in the United
States and Canada over the next five years:
"between fou r bi 11 ion and seven and a hal f
billion dollars".

KEY TO SOME OF THE ABBHEVIATIONS

The total of the figures in Column (6) of
the tabulation, "Likely Spending Next Five
Years", for the 201 cases in the sample, is
"between $415,459,500 and $758,660,000," or
rounding off, "between $400,000,000 and
$750,000,000". In this sample, there are 70
cases not giving any figures; they have been
added as zero; but let us ignore that in order
to be conservative.

acty
admn
anal
apln
chmn
comp
ctr
des
EDPM

It is possible to take this total and
multiply it by a factor F, and arrive at a
rough and approximate estimate of the market
for computer products and services. Everything however depends on the amount of the
factor F.
The factor could be chosen as 10 or II-the ratio of 2000 or 2200 to 200, but there
are many reasons why this is incorrect and
should not be done. (1) It is not necessarily true that the sample here is a representative sample; it is far more likely to be nonrepresentat i ve. (2) Exami ni ng the entries in
the tabulation we can see that there is a heavy
preponderance of replies from buyers of computers and systems. Buyers of components are
hardly represented. Yet in order to make the
computing systems shown above, there will have
to be tremendous purchasing or manufacturing
of components to go into them. This would
tend to increase F. (3) There may be bias
associated with returning the survey form. A
number of subscribers to "Computers and Automation" could have said to themselves "Well,
my recommendations would not affect purchase
--this kind of questionnaire is not for me
to fill out and return". This would tend to
decrease F. And there are many more reasons.
But the factor F is almost certainly not
less than 2. For it would be very hard to
believe that a 9 percent sample has reported
50 percent of the market. Also it would be
hard to believe that the factor F is much
greater than 30; for it would be hard to believe that a 9 percent sample has reported
only 3 percent of the market.
Suppose then that we temporarily apply a
factor of 10, until the time when we have compared these reports more carefully and made
industry by industry predictions. Then we
- 9 -

EE
eng
eqpmt
flt
gp
indir
ins
ldr
leg
meth
oper
orgn
pIng
phys
poss
proc
procg
prom
prgmr
rec
ref
ret;
rep
res
sci
sim
stat
sys
tabg

actuary
administration, -or
analyst, -is
application, -s
chairman of comrr!ittee
computer, -ation
center
design
electronic data processing
machines
electrical engineer, -ing
engineering
equipment
flight
group
indirectly
insurance
leader
legislation
methoQs
operatio.l1s
orga.l1ization
planning
physicist
possibly
procedures
processing
promotion
programmer
records
reference
regional
representative
research
science, -tist
simulation
statistician
systems
tabulating

Computers and Automation

Tabulation 1
ESTIMATE OF COMPUTER MARKET -SUMMARY OF REPLIES MARCH 8 TO APRIL 15

(5)
Products and Services
(c)
(a)
(b)
(d)
(e)
Q!.9.. Anal. Other Compo Servo

(3)

(4)

Size

Recom.,
Title

1.Bl .digital computing service

yes, dept
director

lB2

reliability research

yes, div
mgr

lB3

radar & radar data procg,
special receivers, gen.
electronics

yes, engrg
consltnt

DMT

lB4

insurance

yes, dir
meth res

IB5

air transportation

yes, dir
res

IB6

custom designed processes
and systems

yes,pres

lCl

investment services

yes, div
mgr

lC2

industrial instruments

yes,comptr
supvr

IDI

electrical machinery

yes, -

ID2

air conditioning

yes, dir
of proc

lD3

semiconductor devices

yes, gen
mgr

lEI

oper res services

yes, sr
res assoc

IFI

instruments

yes,staff
engr

IF2

chemicals

yes, engr

lF3

life insurance

yes, sr
res assoc

IG1

industrial instruments

yes, -

IHI

elecnc tube sales dept

IH2

electrical apparatus and
eqpmt

Code

(2)
Type of Organiza,.!.ion

183 chemicals
184

heat transfer apparatus

185 banking

(6)
Likely Spending
Next Five Years
Between:
And:

$ 75,000

$125,000

2,000,000

4,000,000

1,000

500,000

400,000

1,000,000

1,300,000

1,500,000

6,000,000

600,000

700,000

100,000

200,000

10,000,000

15,000,000

2,000,000

3,000,000

300,000

500,00~

S
M

0
P
DMT
v

MTO

25,000,000

35,OOO,QOO

500,000

1,000,000

- , analyst
v

yes,
consltnt

yes, mgr

yes, dept
mgr

250,000

350,000

yes, 2nd

15,000

2,000,000

-10 -

Computer Market
(2)

(1)

(3)

(4)

Size

Recom.,
Code

Type of Organization

(5)
Products and Services
(c)
(a)
(b)
(d)
(e)
Dig. Anal. Other Compo Servo

III

banking services

yes, VP

II2

banking

yes, res
analyst

IJI

research

yes, div
head

IJ2

data procg machines

no,

IJ3

computers, data procg
eqpmt '-..

IKI

mfg surgical dressings

yes, supvr

ILl

consulting service

yes,mgmt
engr

IL2

naval reactors

yes, mgr

IMI

retail chain apparel
men & boys

yes, treas

1M2

detergents, shortening,
marQilrine, dentifrice

INI

banking services

IN2

(6)
Likely Spending
Next Five Years
Between:
And:
$15,000

$ 2,000,000

25,000

250,000

400,000

, 5,000,000

8,000,000

50,000

60,000

yes, sr sys v
analyst

250,000

750,000

yes, asst
cashier

480,000

700,000

drugs, fine chemicals,
pharmaceuticals

yes, dept
mgr

1,000,000

2,000,000

101

naval ship construction

yes, CDR
SC

8,500

60,000

IRI

service orgn (library)

yes, adm
asst

lSI

printing & punched card
services

yes, VP

150,000

200,000

IS2

office eqpmt

IS3

education & research

yes, asst
prof

100,000

500,000

IS4

casualty, fire & marine
insurance

yes, mgr

2,500,000

3,000,000

ITI

electronic research

yes, acct
mgr

2,500

5,000

IT2

oil & chemicals

IT3

engrg development

yes, exec
engr

IT4

gen mail order

yes, comp
ctr mgr

ITS

textiles

yes,

P
SO

nO', applied v
sci rep

reg'l
mgr

yes, asst
mgr

v
M

-11 -

2,500,000

7,500,000

DMI'O

100,000

250,QOO

2,500,000

5,000,000

Computers and Automation

(2)

(1)

Code

T~Ee

of Organization

(3)

(4 )

Size

Recom.,
Title

(5)
Products and Services
(a)
(c)
(e)
(b)
(d)
Qi[. Anal. Other ComE. Servo

lUI

research & development

yes, head
v
math apln gp

lU3

electrical wire & cable

yes, methods v
dir

lVl

elecnc digital computers

- , advt &

(6)
Likely Spending
Next Five Years
Between:
And:
$1,000,000

$ 2,000,000

500,000

1,000,000

200,000

250,000

3,000,000

10,000,000

100,000

500,000

700,000

1,000,000

750,000

1,000,000

600,000

800,000

2,000,000

4,000,000

prom mgr
lV2

textiles

yes, mgr

lV3

maps

yes, sec
chief

lWl

valves

yes, staff
speclst

2Bl

electronic eqpmt

yes, oper
mgr

2B2

rayon, nylon, staple

no, mgr

2B3

ord, A & S group, pension
insurance

yes, asst
VP

2B4

fire & casualty insurance

yes, VP

2Cl

shoe retail & mfg

no, IBM
mgr

2C2

prototypes (miss & flares)

no, anal

2C3

computers

yes, reg
mgr

yes, dir

2C4

201

petroleum

yes, asst
to mgmt

202

aeronautical electronics
systems

yes, VP

203

test & evalln aircraft
arm systems

yes, supvsr
mathn

204

containers

yes, dept
mgr

205

aircraft

yes, desn
speclst'

2El

life, fire & casualty ins

yes, mgr

2Fl

ferro-alloys & metals

yes, mgr

2Gl

aircraft

-, chief

2G2

shoes

yes, tabg
mgr

2G3

computers

yes, project v
supvsr

.2G4

automobiles

yes, anal

v
- 12 -

OMT

PS
S

500,000

1,000,000

100,000

200,000

500,000

625,000

600,000

800,000

500,000

.1,000,000

550,000

700,000

1,200,000

2,500,000

OTO

OMTO

Computer "arket

(2)

(1)

Code

T~~e

of Organization

(3)

(4 )

Size

Recom.,
Title

(5)'
Products and Services
(c)
(a) (b)
(d)
(e)
Q![. Anal. Other Com~. Servo

2G5

steel producer

yes,
supvsr

2Hl

rubber, plastics, etc

yes, mgr
EDP

2H2

AEC installation

yes, comp
supvsr

2Jl

business machines

yes, devel
engr

2Ml

life insurance

no, assoc
acty

2Nl

transformers & components

yes, dir
engrg

2N2

1ibrary service

yes, admn
asst

201

business machines

yes, head
res dept

202

airplane missiles, reactors

203

airframes

204

!j) 875,000

$1,250,000

100,000

200,000

1,000,000

1,500,000

600,000

750,000

MTO

100,000,000

200,000,000

35,000,000

50,000,000

4,620,000

6,200,000

yes, gen
supvsr

yes, EDPM
supvsr

ordinary life insurance

yes, asst
act'y

2PI

research (atomic energy)

yes, mathn

2P2

1 He insurance

yes, asst
compt

2P3

petroleum products

yes, sr
res phys

2Ql

life insurance

yes, data
procg dir

2Q2

data redn systems

yes,

2Q3

pharmaceuticals

yes, mgr
v
mach rec div

2Q4

engrg, consltnt serv

yes, des
engr

251

computers - component

252

digital computers

2S3

steel

2Tl

railroad transportn

(6)
Likely Spending
Next Five Years
Between:
And:

1,800,000

2,800,000

DMTO

P50

5,000,000

7,500,000

400,000

500,000

700,000

240,000

300,000

750,000

1,250,000

80,000

120,000

yes, mfr

2,000,000

2,500,000

- , asst

180,000

200,000

pass,

v
v

0
v

engr vaIn
2T2

savings accounts

- , audit

2T3

basic agric & i ndustl
chemicals

yes, sys
supvsr

- 13 -

Computers and Automation
(2)

(1)

Code

T~~e

of Organization

(3)

(4 )

Size

Recom.,
Title

(5)
Products and Services
(c)
(a)
(b)
(d)
(e)
Dig. Anal. Other Com~. Servo

2T4

dept store

yes, proj
rep

2Ul

pipe forgings, flanges,
fittings

yes, sys
analst

2U2

detect syst transistors
elect comp

yes, data
v
procg supvsr

2Vl

leasing of tank cars

part, asst
s &P

2V2

insurance

yes, const

2V3

insurance

yes, mgr

2V4

admn fed leg re rail
workers

indir,

2V5

medical & hospital care

no,. asst
med admn

2Xl

nuclear consltg services

yes, dept
head

2X2

research & devel

yes, sec
mgr

2X3

digital computing serv

yes, supvsr
engr

2X4

services to medicine

yes, pathol

2X5

statistics

no, prgmr

2Al

radio, TV, hearing aids

yes, supvsr
tab dept

381

rocket engines

yes, sr
engr

382

designs

yes, EE

383

baked goods

384

insurance underwriters

385

$17,500

$225,000

1,000,000

1,500,000

700,000

1,000,000

500,000

1,000,000

1,125,000

1,500,000

300,000

500,000

100,000

500,000

P
v
v

(6)
Likely Spending
Next Five Years
Between:
And:

v
MO

s~pvsr

v
v

MTO

25,000

50,000

2,500,000

3,000,000

2,500,000

5,000,000

1,750,000

2,500,000

125,000

250,000

3,000,000

5,000,000

100,000

300,000

yes, offc
mgr

yes, data
procg anal

data procg for geodetic
div

no, mathn
proj ldr

386

management education

yes, spec
proj dir

3Cl

research & development

3C2

data procg, comptg, math
& stat res, surveys

3C3
3Dl

v
PS

-, mgr math v
anal sec

yes, chief
res sci

airplanes

yes, actg
supvsr

transportation service

yes, sys
dey chief

- 14 -

1,680,000

3,560,000

5,000,000

10, 000, OOu

Computer Market

(2)

(1)

Code

Type of Organization

(3 )

(4 )

Size

Recom.,
Title

302
303

(5)
Products and Services
(a)
(c)
(e)
(b)
(d)
Qi[. Anal. Other Compo Servo

no, sr
prgmr
life insurance

304

partly,
asst mgr

yes, head
v
flt simlab

v
v

2,500,000

OMTO

3,000,000

5,000,000

1,300,000

1,500,000

375,000

500,000

4,000,000

6,000,000

500,000

1,000,000

1,500,000

2,000,000

200,000

3,000,000

225,000

275,000

3,000,000

4,000,000

800,000

900,000

250,000

1,000,000

yes, mgmt
anal

3E2

answers

yes, data
procg dir

3E3

logistics

yes, OEPLOPOO

3E4

research & development

yes, VP
v

2,000,000

$400,000

agriculture

- , ref

$300,000
v

3El

3Fl

(6)
Likely Spending
Next Five Years
Between:
And:

OMTO

OMT

M
M

librarian

3Gl

insurance

yes, mgr

3G2

banking

yes, VP

3G3

aluminum extrusions &
steel sash

yes, compr
mgr

3Hl

none, general composing
function

yes, mgr

3H2

digital computers

yes, lab
adm

OMTO

.3H3

life insurance

yes,
coordtr

3Jl

digital computers

no, sci
rep

3J2

digital computers & acctg
machines

no, assoc
engr

3J3

petroleum products

yes, chmn

3J4

insurance

yes, supvsr
elecncs'

3J5

potentially,
partner

3Ml

electronic

yes, data
proc mgr

3M2

casualty insurance

yes, mgr

3M3

airplanes

yes, mgr

3Nl

ship construction and
repairs

yes, asst
comptrlr

3N2

insurance

yes, supvsr

301

adding machines, cash
registers, data proc sys

no, sr sys
engr

v
v
v

-15 -

v
v

7,000,000

10,000,000

5,000,000

10,000,000

M
MT

Computers and Automation

Code

(3)

(2)

(1)

Tn:!e of Organization

302

aircraft & assoc

303

natural gas transmission

3Pl

Size

(4 )

(5)
Products and Services
(a)
(b)
(c)
(d)
(e)
Dig. Anal. Other Compo Servo

Recom. ,
Title
yes, engr

yes, data
proc dir

crude oil & refined prods

yes", s upvsr v

3Ql

truck transportation

yes, dir
acctg

3Q2

telephone service

3Q3

personnel research

3Q4

~qpm

(6)
Likely Spending
Next Five Years
Between:
And:

DMTO
$2,000,000

$3,000,000

2,500,000

3,000,000

250,000

500,000

no, sr
engr

yes, anal
stat

engines & accessories

yes, gp
Idr

3S1

electronics & research

yes, techl
staff

3S2

research

no, asst
mathn

3S3

research for air force

yes, dept
chief

3S4

research & techl training

yes,

3S5

compr systems

PS
150,000

300,000

400,000

1,000,000

3,000,000

6,000,000

MTO

10,000,000

20,000,000

5,000,000

10,000,000

500,000

5,000,000

50,000

75,000

M
S

- , sys
anal

3T3

state government

yes,·techn
tabg sys

3T4

compensation insurance

yes, div
chief

3T5

petroleum products

yes, chmn

3Ul

computer services

yes, pres

3U2

core memories, matrices,
buffers, data procg sys

yes, VP

3Vl

pharmaceutioo;als

v
yes, asst
chief acctnt

3V2

personnel research

no, psychst v

3Xl

written communications

no, div
rep

4Bl

electronic & nuclear R & D
production

yes, head
data procg

4B2

military & industrial

yes, ohief
engr

4Cl

research

yes, head
comp lab

4C2

airplanes

no, res
engr

S
v

- 16 -

MTO

10,000,000

25,000,000

1,000,000

2,200,000

2,000,000

2,500,000

MTO

3,000,000

5,000,000

DMTO

5,000,000

10,000,000

Computer Market

(2)

(1)

Code
4C3

Type of Organization
electric utility

(3 )

(4 )

Size

Recom.,
Title

4Dl

. (5)
Products and Services
(a)
(b)
(c)
(d)
(e)
Dig. Anal. Other Compo Servo

yes, sys
anal engr

yes, supvsr

(6)
Likely Spending
Next Five Years
Between:
And:
$240,000

$480,000

4D2

aircraft mfg

yes, gp,ldr
compg

600,000

1,500,000

4D3

electrical apparatus

yes, mgr

2,000,000

3,000,000

404

aircraft

no, res
engr

4D5

aircraft

no,
prgmr

50,000,000

100,000,000

4El

technical college

no, ref
libr ' n

4Fl

data procg serv, consltg
serv

yes,
owner

20,000

60,000

4F2

petroleum research

yes, chern
engr

700,000

900,000

4Gl

communication equipment

yes, mgr
purchg

4Hl

computers

yes,

15,000,000

30,000,000

4Kl

el. aircraft power systems

yes, staff
engr

100,000

300,000

4Ll

training of students

yes, head
dept EE

yes, chief
res & pIng

2,500,000

3,100,000

100,000

500,000

15,000,000

20,000 ,000.

4Nl

5
SO

v
v

MTO
M

DT
SO

SO
M

P
P

4N2

credit information

yes, exec
mgr

4N3

business consultants

yes,

401

telephone service

yes,
supvsr

4Pl

instruction, res services

yes, asst
prof

4Ql

banking services

yes, AVP

451

~ata

no, meth
anal

452

antenna couplers, spec
purpose computers

no, serv
engr

453

aircraft components

yes, sys
anal

4Tl

contractor to AEC

yes, div
supvsr

DMTO

PSO

4T2

airborne analog comptr

yes, sec
head'

procg eqpmt

v
v
v

MTO

300,000

600,000

500,000

2,000,000

10,000

40,000

DMT

(cont'd on page 44)

- 17 -

NEW PRO,D,UCTS, AND, ID,EAS
A THREE-AXIS SIMULATOR FOR CONTROLLED FLIGHT TEST OF AIRBORNE SYSTEMS
Bendix Computer Division
Los Angeles, Calif.
A new simulator is resulting in faster
and mor~ efficient testing of controls and
guidance systems for missiles and aircraft.
It is comprised of a 3-axis flight table and
a precise analog simulator control unit; it
reduces or eliminates the need for expensive
trial flights in the evaluation and de-bugging
of airborne systems.

tion absorbing base, can be moved outdoors
for line-of-sight tracking simulation of missile guidance systems.- It produces the motions of roll, pitch and yaw through precision hydraulic servos.
The 3-axis simulator's electronic section contains its own analog computer, designed to exactly match the requirements of
the system. The hydraulic supply is also
included, eliminating the need for auxiliary
movement.

Tests can be conducted with angular motions at accelerations and velocities programmed into the computer. The three gimballed assembly, mounted on a massive, vibra-

The flight table specifications follow:
Maximum acceleration, with
50 lbs. load mounted on
the table, in radians per
second: roll, 2500; yaw,
500; pitch, 500.
Maximum velocity, in radians
per second: roll, 50;
yaw, 15; pitch, 15. Altitude range: continuous
on all gimbals. Frequency
for 90 degree phase shift
(with the same load as
above), in cycles per second: roll, 100; yaw,
45; pitch, 45.

- 18 -

New Products and Ideas

A NEW RADAR DEVICE FOR CORRECTED MAPPING

item sufficiently'interesting for publication
in the "New Products" section of your magazine.

Fairchild Controls Corp.
Syosset, Long Island, N.Y.
A novel device which promises to inaugurate a new era in the use of radar for accurate mapping and charting has been completed.
The instrument has been named the "Radar
Restitutor"; it was invented, designed and
fabricated by the Electronics Division of
Fairchild Controls Corporation, a subsidiary
of Fairchild Camera and Instrument Corporation.
Photographs of radar displays are not
normally true maps since they contain many
distortions and inaccuracies. This makes such
photographs unsuitable for measuring distances
between different points on the ground. The
Radar Restitutor automatically computes the
errors which give rise to the distortions
present in the radar photography, and it produces a new photograph on which every ground
object is accurately located. The photographs
produced by the Restitutor are then used in
the compilation of maps.
The chief distortion on the original
radar photographs is caused by the fact that
radar records the distance between the radar
station and an object, and for airborne radar
sets this means that the slant distance is
recorded. For mapping purposes, though, it
is necessary to know the ground distance between the point directly under the aircraft
and the ground object. Another source of
error is the fact that an airborne radar is
continually moving, while the radar display
does not take this motion into consideration.
The Restitutor not only computes these distortions but several additional distortions
as well, and positions the grou~d object at
the corrected coordinates on the new photograph which it produces.

These Lattice Networks for digital storage consist of a novel type of lumped constant
electromagnetic network specifically developed
and designed for digital storage. It uses
only 1.2 elements per bit, and provides highly
stable characteristics without extreme tolerances in the individual elements.
Hermetically sealed units can be supplied
for any capacity up to 30 bits at digit rates
up to 5 megacycles, for satisfactory operation
over a temperature range of - 20 0 C to t 70 0 C.
The units provide an extremely economical and
compact form of storage in this capacity range,
and can readily be grouped for serial or
parallel operation with rapid access.

*-------------------*--------------------*
A NEW IDEA IN COMPONENT SUPPLY

v. Guignon
Great Neck, N.Y.

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

In 1952, a new idea in supply of electronic components was started. The idea was
simple but had never been tried; it was to
gather under one roof a maximum of ten important lines of components -- no more --,to know
everything possible about those ten lines, and
to stock amply every item manufactured within
those lines. Users could then count on sufficiently large stock in the distributor's hands
to have their orders delivered to them in hours
instead of weeks. Any engineer in charge of
producing equipment to meet a deadline knows
well that a bottleneck in supply is just as serious as any other kind.

LATTICE NETWORKS FOR DIGITAL STORAGE
R. H. Davies
Ferranti Electric Inc.
New York 20, N.Y.

To the Editor:
Following is a brief description and a
photograph of a new type of lattice network
which we have developed for use in digital
memory stores. We trust 'you will. find this

Now, in 1957, the company, Schweber Electronics, Mineola, N.Y., has grown to 30 persons,
(cont' d on page 41)

-19 -

RELIABILITY
BUSINESS

AUTOMATIC

DATA

PROCESSING

Herbert T. Glantz
John Diebold & Associates, Inc.
New York, N. Y.
(Based on a talk delivered February 27, 1957 at the Western Joint Computer Conference
Los Angeles, California)
Introduction
Throughout the past few years a great
deal of study has been devoted to analyzing
the different characteristics and requirements of scientific and commercial data processing systems. Although early general
agreement was reached on the fact that such
systems were different, the exact cause and
nature of these variations has not yet been
clearly defined. Equipment manufacturers
attempted to resolve this difficulty by designing two distinct "lines" of computer
models. In some instances these manufacturing distinLtions have become blurred with
usage, with the result that a Remington Rand
Univac is utilized for engineering calculations, while an IBM 704 is applied to payroll
preparation. But, in the main, this dichotomy of design and application is being effectively preserved.
Dr. Jay Forrester, the former Director
of MIT Project Whirlwind, has characterized
the chronology of electronic computation as
falling into three distinct phases. Beginning in 1945, an intensive amount of research
was devoted to investigation of the physical
possibility of building electronic digital
computers. The early studies of computer.
logic and circuit design were conducted almost entirely by various engineering universities. By 1950 it was apparent that the
basic problems could be solved and that electronic computation would become a reality.
Shortly afterwards, the major portion of research activity was shifted to the application of these machines to various problems.
The second phase of the history of electronic
computation was devoted to computation in the
fields of engineering and science, as the
first large-scale digital computers began to
appear in the universities and aircraft companies. In 1955 the emphasis of machine applications research began to switch from
engineering to commercial problems. This
trend marked the entry into the third and
perhaps the most vital phase of development
of electronic computation.
As the problem of utilizing electronic
- 20 -

equipment for business applications was approached, it became apparent 'that there were
major differences in the requirements of this
new area. The early studies devoted to this
problem concentrated on seemingly obvious
operational variations. Business problems
called for vast amounts of input-output data,
while scientific problems required lengthy
and involved internal calculations. Thus,
in short order a convenient categorization
grew up:
Engineering and Scientific Applications:
Small volume of input-output data; large
amounts of complex internal calculations.
Commercial Applications: Large volumes
of input-output data; small amounts of
simple internal calculations.
Although this fairly arbitrary and sweeping
.classification proved adequate for early needs
of the industry, an increasing sophistication
in computer usage has tended to obscure these
demarcation lines. It has become increasingly
evident that such operational differences do
not provide an adequate representation of the
two different systems.
If one considers instead the respective
functional purposes of scientific and commercial systems, a striking contrast may be observed. Scientific data processing installations may be regarded as self-contained systems which function only to satisfy the dictates and requirements of the parent organization. Thus, in a broad sense, all input data
originates within the system proper; all computations are determined by the needs of the
overall organization; and the timing, quantity,
and amount of output information is again dictated by the requirements of the system itselL
Commercial data processing systems, on
the other hand, must inevitably exist and perform their functions in constant relationship
with the environment of the business world at
large. This relationship, which dominates
both the design and performance of business
systems, is notably evidenced in three ways.
The mass of data entering the system originates externally and is to a large extent
uncontrolled in format, timing, content, and
accuracy. The calculations that are performed

Reliability
are frequently regulated by the rules of outside agencies such as the S.E.C., I.C.C., and
the Bureau of Internal Revenue. Finally, the
daily deadlines that must be satisfied are
generally determined by an essentially indifferent environment and are frequently unyielding and seemingly unrealistic.
Accordingly, whereas a scientific data
processing system exists and operates to satisfy its own n~eds, commercial systems must function in large measure to satisfy requirements
imposed from outside the system.
The Design of Business Systems
Business systems are growing increasingly
more complex as their range of applications
expands to cover more demanding and more intricate areas. A functioning commercial data
processing system includes a variety of information handling components, human, mechanical, and electronic, which are linked together
by an overall communications network. The
term "integrated data processing" has gradually corne to be accepted as a generic description of the workings of such systems.
The design of these systems has created
a strikillg opportunity for the business world.
We are now able to conceive of all routine
daily operations of a company being controlled
and directed by an automatic system. Automatic process control is already a familiar
occurrence in the chemical and petroleum industries. We are on the verge of applying
similar concepts to a number of commercial
enterprises. The underlying theme in such
systems is automatic control of business
operations including a feedback loop for the
correction of errors. At the same time, the
managing executive is to be provided with
information that will allow him to exercise
control over entire company policy on the
basis of reliable and timely data.
However, in painting this picture of
successful office automation we have passed
over a number of evidently dangerous pitfalls.
For as automatic systems exercise a greater
amount of control and direction, one must
place a correspondingly greater emphasis on
their operating reliability. Intermittent
or periodic failures in such systems can
wreak more havoc than is to be gained from
long intervals of reliable operation. The
increased complexity of these systems introduces more components that are liable to failure and, due to their varying interrelationships, introduces a greater degree of difficulty in isolating and replacing the faulty
elements. Many systems engineers become so
enamored of possible accomplishments that
they tend to overlook the question of system
- 21 -

reliability and the implications of component
failure.
As an illustration of the problems that
are inherent in the operation of business data
processing systems, I should like to briefly
discuss two systems that our organization has
designed for commercial organizations. In
both cases we have completely restructured the
data processing operations of the companies
involved. In both cases, although substantial
dollar savings were accomplished, our primary
aim was to have the data processing system
provide more effective aid to the sales organization and to top management of the company.
In both cases this was effected by centralizing the data processing elements and by utilizing extensive communications networks.
A Stock Brokerage House
"X" is one of the largest stock brokerage
houses in the country. Their main volume of
business originates on the West Coast where
they maintain a number of branch offices. One
of the principal requirements of competitive
existence in the brokerage field is rapid fulfillment of customers' orders, and requests
for sample prices, or "quotes". Since most
of the clients' trading is done on the floor
of the New York Stock Exchange, the company
utilizes an extensive private wire system to
link the branches with the Eastern trading
center. It is not unusual for an order to
originate in Los Angeles, be flashed to New
York, be executed on the Exchange, and be confirmed back to Los Angeles in a matter of two
or three minutes.
The system that we have designed for this
firm includes an automatic teletype switching
center on the West Coast and a Data Processing
Center in New York. The data processing center utilizes a medium-scale general purpose
digital computer as the nucleus of all information-processing activities. All branch
orders are received by the West Coast message
center and are automatically routed on to New
York where they are simultaneously sent to
the Exchange and to the data processing center. Execution reports from the Exchange are
sent directly to the originating branch and to
the data processing center. These orders and
execution reports are received on 5-channel
teletype tape which is fed directly into the
computer. The volume of business handled
during the five and one half hour trading
day, coupled with overall speed requirements,
necessitate a system capacity of twelve such
order-execution pairs per minute.
Basic input is also provided to the system by the branch and New York office reports
of daily receipts and disbursements of cash
and securities. This extensive input data

· Computers and Automation

must be processed each day according to a
rigid timetable. In general, all of one day's
business activity must be completely recorded
and processed before the start of the next
trading day. All of this work is subject to
the detailed scrutiny of various Exchanges,
the S.E.C., and the auditors.

ating nature of data transmission time schedules, the overall system design is based on
combined use of direct teletype input and
output and extensive air mail communication.
Original Shipping Orders, as well ~s other
operational data, are sent to the data processing center, which functions as the controlling element for the entire company. In
addition to processing routine daily operating
data, the computer is utilized for the formulation of strategic decisions as to selection
of optimum freight car routings and CarrierTariff combinations.

By utilizing the extensive private wire
system, the data processing center in New
York is able to direct the complex daily
operational activities of this company.
Furthermore, although the company's top management is separated from the data processing
center by 3,000 miles, the center is able to
provide them with a normal supply of reports
while at the same time rapidly fulfilling
requests for special analysis.

Operational

A Transportation Company
"Y" is a medium-size company in the transportation industry. The main volume of their
work is concerned with cross-country movement
of railroad freight cars. The majority of
such movements originate in East Coast ports
and are destined for Western cities. However,
a considerable portion of volume consists of
overnight movements into such mid~~stern
cities as Chicago, St. Louis, and Cincinnati.
A further complication is caused by the frequent necessity of transfer operations, as
when Baltimore and Boston shipments are merged
in St. Louis before going on to Los Angeles.
A Shipping Order is prepared by the customer and delivered to the Company with each
consignment of merchandise. These Shipping
Orders are the basic input to the system and
are used to prepare Freight Bills, Railroad
Manifests, Receivabl~s Entries, and so forth.
A completed Freight Bill must precede each
shipment to its destination and must be on
hand at the appropriate Transfer Points before arrival of the various freight cars. All
daily processing activities are subject to
I.C.C. regulation and are continually compared by Carrier Railroads with their own
computations. Our client processes an average of 10,000 Shipping Orders each working
day.

Characte~istics

The X Brokerage House and the Y Transportation Company are in two vastly different
fields of business. Yet the requirements and
characteristics of the data processing systems
that have been designed for these companies
are strikingly similar. These similarities
reveal a great deal about the nature of such
business systems:
Both companies operate over a wide geographic area and use an extensive communications network and a medium-scale
computer in a central data processing
installation.
In both cases, basic input data is provided by activities that are outside of
the system and are essentially uninterested in the workings of the system
and the difficulties that are caused by
incorrect data.
In both cases, input data arrives in a
fairly random fashion but must nevertheless be processed upon receipt, since
all operations are conducted against
fairly intractable time deadlines.
In both cases, normal operational schedules must be flexible enough to accommodate the intermittent interruptions
of rush jobs.
In both cases, internal system processing must conform to rulings of various
outside regulatory agencies.
"Real Time" Business Systems
The most important characteristic of both
systems, however, is that they are "real time"
business systems. The standard definition of
a real time data processing system.is one
"whose actions influence the input data that
is being received". In the world of business
data processing systems, this definition may
be modified to read: "A real time business
system operates on-line with its input data".
One such automatic real time business system
is the American Airlines Reservisor system
wherein passenger requests for seats are
satisfied as they are received with essentially no processing delay.

The data processing operations of this
company are characterized by a large mass of
input data, a requirement for flexible processing schedules that allow for the periodic
interruptions of rush movements, and a high
volume of printed output accompani~d by a
moderate amount of punched paper tape output
suitable for direct teletype transmission.
Our system design for this company utilizes a medium-scale general purpose digital
computer installed in a Data Processing Center located in Chicago. Due to the fluctu- 22 -

Reliability
Until very recently, almost all commercial installations of general purpose computer
sys~ems were used in non-real time situations
and relied extensively on "batch processing"
techniques. Examples of areas that are amenable to such techniques form an honor roll of
the problems first handled by business systems: Payroll, Receivables Accounting, Inventory Accounting, Insurance and Utility
Billing, and so forth.
'
The functional nature of a real time
business system requires that essential control and direction of the business be vested
in the data processing system itself. Both
of the systems described above receive information concerning the environment or stimuli or' the business. In one case these are
orders for security trades; in the other, they
are Shipping Orders. The automatic system
processes the information contained in these
initial messages and issues directions so
that successive appropriate measures may be
taken. Finally, the data processing 'center
is notified of the results of these actions
and acts to issue either corrections or instructions for further measures. All operational data funnels into the processing system and all routine operational directions
issue from the system. Further, the formulations of company management are based on
analysis prepared by the data processing
system from information that is contained in
the records of the system.
An operation of this type sounds delightful to the systems engineer, for we have in
large measure eliminated the human element
from routine business functions. The automatic system directs and controls daily operations while using the results to provide
management with timely reports on which to
base long-range policy decisions.
However, the practical businessman views
such a system in an entirely different light.
The system will reduce direct operating costs;
it will provide greater flexibility and efficiency than the present manual system; management will receive information in time to formulate important decisions that direct the
company's future operations. But, while accepting the validity of such advantages, the
executive also realizes that if this beautiful system should fail, his company will be
out of business.
In the final analysis, this factor of
reliability is the most important characteristic of real time business systems and provides the greatest single functional difference between on-line and batch processing
systems. If the payroll is late, there will
be unhappy labor force, and if the utility
bills are delayed the company will be corres- 23 -

pondingly tardy in receiving revenue. But if
an on-line s~stem fails, all company operations cease.
Reliability in Real Time Systems
At the same time, the inherent benefits
that may accrue from the utilization of automatic real time business systems are significantly greater than those provided by the
earlier systems. It is just this factor of
automatic control of routine operations that
provides the appeal of on-line systems. For
the effect of a day's activity will be reflected in reports early enough for the executive to exert an effective influence on
these same operations. Such real time data
processing systems are capable of implementing the theory of "management by exception"
in actual daily operations.
We thus have a system concept that can
provide extensive and valuable benefits to
a business organization and can also do irreparable harm in case of failure. Accordingly, we must provide a system design that
will reliably ensure against failure while
retaining the maximum benefits. Only in this
fashion, can we hope to implement this significant step forward in the application of
automatic data processing.
In the design of real time business systems we have adopted a concept that is based
on the assumption that all components of a
system are liable to failure, but that overall operations must not fail. One of the
simplest and most effective means of ensuring
system reliability is through the use of redundancy. This technique is utilized by the
human brain and nervous system, and the analogy is a natural one since the data processing
system functions in a similar fashion for the
business organism. However, in view of the
economic realities of commercial life, it
would appear more practicable to approach the
problem in a slightly different manner.
The first requisite of our system is
high quality of design and performance in
each individual element. Practically, this
implies that only proven components can be
incorporated in the system. As newer computers are developed, they should be "broken
in" on the batch processing problems. If
that is not feasible, one must insist on a
substantial period of rigorous testing before
accepting such equipment.
Continual preventive maintenance of all
equipment is required. A number of engineering installations of medium-scale computers
adopted the practice of eliminating preventive maintenance periods. These organizations
(cont'd on page 42)

AUTOMATIC

LIBRARY

DOCUMENTS

Richard Moyer
University of Pennsylvania, The Institute for Cooperative llesearch
Philadelphia, Pa.

(This work was supported in full by the Air Research and Development Command under Contract
No. AFI8(600)-125 with the Institute)
The idea of an automatic library has been
the object of speculation for a considerable
length of time. Various schemes, some grandiose and some quite modest, have been advanced
for its accomplishment. Some involve machinery that would stagger the designers of a
modern oil refinery, while others are quite
simple; some display Rube Goldberg overtones,
and others seem quite sensible. Most of them
aim at a level of automation and a speed of
operation that still require development and
research before becoming realities. However,
if these schemes are discarded and an examination is made of the various types of automatic machinery already in production in the
plants of electronic computer manufacturers,
it will be possible to automatize many of the
operations of the modern library. Although
the ultimate degree of automation has not yet
been realized, sufficient progress has been
made to eliminate many man-hours of labor.
While the possible speed of operation is not
phenomenal, when viewed in relation to the
speed of present methods it does represent a
very substantial saving of time. The cost,
furthermore, will be considerably less than
that for present methods of operation.
Currently, both industry and science are
finding it increasingly difficult to retrieve,
for research purposes, information stored in
special and general purpose libra~y collection~
This situation is the result of the present
overflow of information that not only contributes to the time and care required in finding anything specific in an already massive
collection, but also complicates the task of
adding the constantly expanding incoming flow
of new information to that collection.
Naturally, a great capital investment is
necessary to install electronic computers and
automatic maChinery in a library. This fact
may be viewed by many with trepidation, but
several industries and government agencies
already have electronic computers that could
be devoted to part-time library work, thus
greatly reducing the initial investment.
Such agencies normally represent a specialized
field of endeavor and have libraries that
might contain 100,000 to 200,000 documents,
i.e., publications or parts of publications
considered as single units for indexing pur-,

poses. If ~n automatic installation were made,
various possibilities of improved operation
would open up, and substantial savings of time
and money, as well as increased thoroughness,
would accrue to industrial research. Electronic
computer manufacturers, consequently, would
soon begin to design and build special purpose
machines for increasing the speed of operation
to a point where the goals of the visionaries
would be approached.
At the present time, the speed of existing production models of automatic data-handling machinery probably makes the operation
impractical for libraries of more than
1,000,000 to 2,000,000 documents, unless the
required machinery is installed in multiple
units. Furthermore, the initial cost of such
an installation would be prohibitive. Although obsolescence might be a serious factor
initially, stimulated interest would soon produce special equipment capable of greater
speed, and such institutions as the Library
of Congress, as well as the predicted
50,000,000 document library of the future,
could be automatized.
The idea of using electronic computers
for searching the document collection of a
library is not new. In 1951 Philip Rutherford
Bagley prepared at the Massachusetts Institute
of Technology a master's thesis entitled Electronic Digital Machines for High Speed Information SearChing.' Because of the rapid growth
of the mass of published articles, books, etc.,
during the past few years, he specified that
the machines must be able to search 50,000,000
documents within a reasonable length of time.
Bagley's method of recording the necessary
information in the memory system of the electronic computer involves the use of a series
of terms called descriptors which consist of
single words used singly or in groups to describe the characteristics of the subject
matter of a document. This information is
arranged in the memory system of the computer
in the form of consecutive lists of descriptors, each list describing a single document
headed by its serial number. This document
record consists of only two types of information: (1) the serial number of a document
which allows for its positive identification;
and (2) a number of descriptors that describe

- 24 -

Incominqfll,.
~"""'\'\'\'\.~
Document!'

DOCOMENT
RECEPTION

Assignment Of
Security
Classification
Assignment Of
Serial Number
Numbering

.J'

,II
Outgoing
for
Orders
..tI
fcS.\\\\\\..\. \. \..\..\.. \.~
~

Documents

Input To Document Collecti,on

....

I
I

DOCUMENT
COLLECTION

"
,II I~

AUXILIARY
CARD

;
"-

JI'

CODING OF
BIBLIOGRAPHICAL
FACTS &
DOCUMENT
DESmIPTION
PREVIOUS TO
FEEDING TO
COMPUTER
MEMORY

.-~\

SEARCH
J~

"'"
A schematic diagram of
various operations and
utilizes an electronic
(information retrieval)

E-t

0::

Co!)

~ ~
::>
0
0

I--

MECHANISM

0::

CI)

COMPUTER

E-'

~
~

"II'

PROGRAMMING
COMPUTER
SEARCHES

~ ~)6~
c..

COMPUTER
MEIDRY

",
"
"-

0::

E-t

I
I
Input To
I
I+- Compu ter Memory--+ I

CATALOG

~
H

DOCUMENT
ACQUISITION

~
~
u

Cataloging For
Auxiliary Card
Catalog
Indexing To
Required Depth

...J
0

E-t

, I
I
~ I

DOCUMENT
PROCESSING

"""":-

...

Outgoing

....
,

"
~

the flow of information among the
parts of a modern library that
computer for literature searches
•

0::
0

t..

~~~

H~f5

Bibliographies

@:~t..
O::~

Incoming

~~ffi
0::

0::

~"""SJ
Requests

Computers and Automation

the contents of the text of the document.
Bagley's method provides for an average of
30.0 descriptors per document, as well as
for a search procedure designed so that various numbers of descriptors in various combinations can be specified as the subjects of
the information retrieval operation. Invest·igation revealed that the digital computer,
Whirlwind I, could search approximately 11,990
documents per hour and would require approximately 41,700 hours to search a collection of
50,000,000 documents. Bagley then outlined an
idea for a 'special purpose computer that
could be designed to search documents at the
rate of 4.66 million per hour (or 10.7 hours
to search a document collection of 50,000,000).
Recently, Harley E. Tillitt, at the U.S.
Naval Ordnance Test Station at China Lake,
California, published a paper entitled An
Experiment In Information Searching Wit~The '
701 Calculator. Like Bagley, Tillitt used in
his experiment the mechanism of the descriptor to delineate the contents of the document,
but he recorded each descriptor only once and
associated serial numbers with it. Thus the
record in the memory system contains lists of
serial numbers of many different documents
headed by a single descriptor, signifying that
each document in the list contains the information labeled by a descriptor at the head of
the list. In this system, each consecutive
list refers to many documents. As in Bagley's
method, the record consists of only two types
of information: (1) a serial number that
allows for positive identification of the
documentj and (2) a number of descriptors
that define the contents of the text of the
document. Tillitt's operation actually adapted
the Uniterm system to an IBM 701 Electronic
Data Processing Machine and used an average
of 7.6 descriptors per document. This method
provides for searches by the IBM 701 of various numbers of descriptors in different combinations.
Using this system, Tillitt found that he
could record the information for approximately
34,000 documents on a 1200 ft. magnetic tape.
Because of divergent factors peculiar to different types of searches, he found that the
minimum search time for 34,000 documents was
20 seconds and the maximum 4 minutes. If the
maximum search time is used as a base, Tillitt's
method would provide a search rate of 510,000
documents per hour, or 98.04 hours for
50,000,000 documents.
Both Bagley's thesis and Tillitt's operation were aimed at overcoming the problem of
retrieving information stored in a document
collection. The solution of this problem
would remove one of the major obstacles in
carrying out adequate literature research today. However, the problems of the modern
- 26 -

library do not consist solely of information
retrieval. Another major obstacle is the
processing of incoming documents before they
are put into the document collection and thence
become available to individuals engaged in research. The processing of documents consists
of assigning serial numbers, numbering the
document, cataloging the document, and preparing appropriate records for document control. With the increasing mass of publications in recent years, processing of documents
has created a serious delay in the progress of
information storage in itself, so that it is
not uncommon for large libraries of one to
two million items to have documents backlogged
for as much as one or two years. Such a situation impedes research, inasmuch as current
publications are not available in collections.
Naturally, this is an important factor, considering the current advances in scientific
research. In view of this fact, it would
seem that there is a need, not only to provide faster means of searching document collections, but also to streamline and mechanize
the processing of documents for a collection.
These things must be done before the problems
of the modern library will be solved even
partially.
A study of the modern library reveals that
the use of electronic computers for search activities will not permit us to eliminate the
card catalog completely. Search activities
which involve checking for the presence or
absence of a particular document, locating a
document by'a given author, finding any document under certain broad classifications, etc.,
can still be carried out best by the use of a
card catalog. However, the card catalog thus
actually becomes an auxiliary aid and should
be considerably streamlined so that it would
contain only author and title cards and a
maximum of three of the traditional sUbject
cards per document.
A new operation, indexing by the use of
descriptors, takes the place of extensive
cataloging. The product of this process is
the record which is searched by the computer
mechanism. The use of descriptors for indexing eliminates the necessity for making numerous involved judgments which are critical to
the efficient functioning of the standard card
catalog as an information retrieval system.
This being true, personnel requirements for
the indexing operation may be somewhat lowered
in respect to the extent of time required for
training, and the level of intellectual ability
of the trainee. This would tend to ease the
existing shortage of professional librarians
and would also release such people for the
more critical tasks of documentation in the
automatic library.
Other items necessary to the functioning'

Automatic Search

of a modern library are the accession record
( a record of each document in the collection
in the order of its receipt), the shelf list
(a record of the shelf position of each document in the collection), and a series record
(a record which allows for the location or
grouping of any given series of documents
issued periodically). These three records
taken together constitute the document control system and they must be produced when
the document is processed.
The author and Mrs. Gwendolyn M. Bedford,
members of the Institute for Cooperative Research of the University of Pennsylvania,
working in conjunction with Dr. Donald Thomsen
and Mr. David Lawrence of the Applied Science
Division of the Philadelphia offices of the
International Business Machines Corporation,
have designed a system that utilizes an IBM
705 Electronic Data Processing Machine to
search document collections. The selection
of IBM equipment for study as to the possibilities of such an application evolved from a
survey of the Sharp & Dohme punch card information retrieval system at West Point, Pa.
This system employs an IBM 101 Electronic
Statistical Machine.
There is no doubt that the system devised
at the Institution for Cooperative Research
is compatible with other automatic machinery,
since other computer builders have in production machines of similar scope and capacity. The intent in this system is to use a
special assembly of IBM typewriter card punch
equipment (or comparable equipment from any
other manufacturer), including a number of
remotely controlled typewriters, to produce
in one operation all the cards for the aux~
iliary card file and all the necessary control records. This assembly of equipment
will produce simultaneously all the necessary
input to the computer memory.
The documentary information fed into the
computer memory is called the document record
in this system. It consists of two sections,
a bibliographical facts section and descriptor section. The bibliographical facts section contains seventeen facts about the document, such as an identifying and locating
number, the author or authors, the publisher,
the date of publication, etc. The descriptor
section of the document record contains an
average of thirty descriptors per document to
define the contents of the document. ' On the
basis of past experience, it was decided at
the time this system was designed that an
average of thirty descriptors per document
is necessary to define the contents completely
enough to locate the various types of information a document may contain. Within the
descriptor section it will be possible, if
it is so desired, to signify such things as
- 27 -

the year or era, the geographic and political
areas to which the contents pertain, and the
basic nature of the contents.
In this system, as in Bagley's, each consecutive list of information in the memory
system of the computer refers to a single
document. The list contains (1) the serial
number of the document, allowing for its
positive identification, (2) the shelf location, (3) the bibliographical facts such as
author, publisher, date, etc., and (4) the
list of descriptors pertaining to that document.
The coding system designed for the descriptor section obviates some of the difficulties existing in earlier systems such as
Uniterm and various other adaptations of
Coordinate Indexing. The system under discussion provides for the division of the descriptors of any given document into arbitrary groups. Thus two or more discussions
in a single document may be specified as
related or unrelated. If this feature is
used in conjunction with certain types of
search programs, the breadth as well as the
level of abstraction of the retrieved information may be controlled. At the option
of the user, this system also provides for
the qualification of descriptors that appear
in the descriptor section of the document.
The system under discussion carries much
more information in the computer memory (the
document record) than the systems of either
Bagley or Tillitt. The inclusion of seventeen bibliographical facts permits one to
locate any document that contains the necessary information, whether the search be
carried out by author, by issuing agency, or
by publication date. The reader can easily
imagine many other combinations that might
be desired. The experience of research libraries, moreover, corroborates the necessity
for such flexibility and' capacity.
Such a document record, with seventeen
bibliographical facts and thirty terms in
the descriptor section, will be stored on
reels of magnetic tape. Each 2500 ft. reel
of tape will hold the record of 14,000 documents. This document record can be searched
by an IBM 705 Electronic Data Processing Machine at the rate of 150,000 documents per
hour. The results of the search will be delivered by the printer of the Elec~ronic Data
Processing Machine in the form of a list
(bibliography) of identifying and locating
numbers for the documents containing the
desired information. Upon request, the
printer could print out the entire document
record of each document. To study the bibliography more intensively before consulting
-the documents themselves, one would need only

Computers and Automation

to decode the document record. One IBM 705
Electronic Data Processing Machine would require 333.33 hours to search a library of
50,000,000 documents (Bagley's goal), using
the system described here. Since such a
length of time is prohibitive, it would seem
that search of a 50,000,000 document library
will have to await the development of some
such device as Bagley's "special purpose
computer".

It should also be noted that, in the areas
that were thoroughly combed, many irrelevant
references were undoubtedly tagged for examination. Although these two situations may
occur in searches conducted by electronic computer, they are not likely to do so because
the descriptor concept of indexing is more
flexible and at the same time more precise
than the conventional method.

The system described here is slower than
Tillitt's operation with the 701 but, were
Tillitt to raise his average descriptor load
from 7.6 to 30.0 descriptors per document, it
appears that his search speed would be cut
from 510,000 to 146,166 documents per hour.
If he were also to include 17 bibliographical
facts, the speed seems to drop to 93,279 documents per hour. This is only a tentative
estimate because there may be several ways of
arranging the record in the memory system of
the computer. Some of these arrangements
might increase the speed of search to more
than 93,279 documents per hour. On the other
hand, there is a chance that it may be impossible to include more than one bibliographical fact in Tillitt's system (in this
case an identifying number or serial number
of the document) because of the way he lists
his records of documents in the computer
memory.
During the winter of 1953-54, the writer
was concerned with a literature search of a
library of approximately 1,000,000 documents
for some important key facts connected with
a problem. After 600 hours of labor had been
expended at the cost of approximately $1425,
nearly 1700 promising document titles had
been assembled from the card catalog, and
from various abstracts, footnotes, indices,
bibliographies, etc. The 1700 documents
themselves had not been examined. For various reasons, including cost and lack of personnel, the search was abandoned at this point.
However, as a conservative estimate, the time
required to make an adequate examination of
the 1700 documents would certainly be in excess of 850 hours. The labor charges for
this operation would have amounted to at
least $2025. At this,point, the costs are
already $3450, without the inclusion of any
charges for stenographic services or materials.
Even with such an expenditure, the search
would have exploited neither the major portion of the collection's periodical holdings
nor a number of other possible source areas.
More important than the fact that some
areas were not searched is the fact that, because of conventional methods of cataloging,
many relevant references were undoubtedly
missed in the areas that had been covered.
- 28 -

What about cost and time outlays for information retrieval procedures of the system
outlined in this article? An IBM 705 Electronic Data Processing Machine costs approximately $28,180 per month. A crew of six operating 21 hours a day, or two people on each
shift, would cost approximately $3000 per
month. Each shift would consist of an operating specialist ($8000 per year) and a service clerk ($4000 per year). It is difficult
to estimate what other personnel and material
are necessary or at what point cataloging and
library control services leave off and literature searching activities begin, but it would
seem that an additional $10,000 per month
would take care of this overlap, as well as
of additional staff and material needs. This
staff would program searches during the day
for the around-the-clock operation and enter
records of documents in the memory system.
Thus the total operating costs add up to
approximately $41,180 per month, or $2059
per day. Under the system outlined, the
machinery could search, barring breakdown and
excluding multiple searches, 3.15 million
documents every 24 hours (21 hours of work).
If the machine were operated 20 days per
month, the cost would be approximately $653.65
per search of 1,000,000 documents, or 18.9%
of the $3450 necessary to carry out the same
operation by a completely human agency. The
result of the search would be delivered in
the form of serial numbers of documents (and
in many cases page numbers of ·the desired
information) which would permit the location
of any document in the document collection.
So far as time outlays are concerned,
such a system would allow the search of a
million documents every 6.66 hours without
an increase in capital investment for machinery, barring breakdowns and multiple
searches. If demands were so heavy as to
make it necessary to install two electronic
computers, search time per million documents
could be cut to 3.33 hours at approximately
the same cost. If all previous data set
forth ,in thi s di scus s ion are accepted, it
can be readily estimated that an equivalent
but less precise and complete search of a
million documents could be carried out by a
completely human agency in 6.66 hours, with
approximately 221 individuals using existing
library facilities and equipment. However,

Automatic Search

the point must be made that the coordination
of a force of 221 individuals to prosecute
such a search represents an almost insurmountable proulem. In addition, it must be
pointed out that the task of procurement of
so large a group capable of pursuing such a
search is virtually impossible. It is certain, in any case, that the cost would still
exceed $3450.

documents searched by a completely human agency
will number far fewer than 6897 per hour.
On the other hand, it can be stated that
the probabilities that the actual operating
speed will approximate the estimates for Whirlwind I, the IBM 705 system, and Bagley's
"Special purpose computer" are high if the
planned procedures are used. Of course, there
is no question about the speed of operation of
the IBM 701 system using TillittVs system,
since his data are drawn from actual experienca

Table 1 (see below) summarizes the speeds
of the various methods of literature searching
that have been discussed in this article.
It must be re-emphasized,that the cost
estimates given here for a human agencyVs
searching operations are extremely conservative. In addition, attention should be called
to the fact that 6897 documents are not actually examined per hour in such an operation.
The major portion of the 6897 are really
documents whose titles, contents, and catalog
cards are not examined or even known by the
searcher. Whereas the electronic computer
actually searches every document record, the
human agency searches only in certain areas
of the card catalog where subject headings,
etc., indicate that the documents containing
the required information are likely to be
listed. Thus, vast areas of the catalog are
eliminated without ever being touched. The
figure of 6897 documents per hour may also vary
widely from searcher to searcher and from
search to search because of differences in
card catalog construction as well as in the
efficiency and thoroughness of the searcher.
In most cases it is highly probable that the

In conclusion, the savings accrued through
the use of automatic machinery versus the completely human agency, insofar as they are predictable at this time, can be summarized as
follows. The IBM 705 system, staffed by the
necessary crew, could do a literature search
of a document collection in approximately
two percent of the man-hours required by a
completely human agency; the cost of such a
search would be approximately 18.9% of the
cost of carrying out the same search by a
completely human agency; and the results of
the search would be more accurate, thorough,
and complete than the results of a search
carried out by a human agency. In respect to
input to the document collection ~~, it
would seem that this can be simplified and
expedited considerably by the use of descriptor indexing in conjunction with existing
items of automatic machinery. Finally, the
electronic computer as an information storage
mechanism offers unlimited possibilities for
a higher potential number of descriptors and
bibliographical facts per document than any
existing automatic equipment.

TABLE 1
Comparative Speeds of Various Search Agencies

Time Required to Search
Searching
Agency
Completely Human
Agency

Documents
Searched
Per Hour
6,897

Whirlwind I

11990.4

IBM 705 system
as described in
this article

150,000

IBM 701 system
using Tillitt's
method
Bagley's Special
Purpose Computer

100,000
Documents

1,000,000
Documents

50,000,000
Documents

145 man hours

1450 man hours

72,500 man hours

834 hours

41,700 hours

.66 hours

6.66 hours

333.33 hours

510,000

.196 hours

1.96 hours

98.04 hours

4,761,900

.021 hours

.214 hours

10.72 hours

83.4 hours

- 29 -

IRE NATIONAL CONVENTION, MARCH, 1957, NEW ,YORK
TITLES AND ABSTRACTS OF PAPERS BEARING ON
COMPUTERS AND DATA PROCESSORS

sequence of numbers at its output. The implementation of the computer may be completely analog or
digital, or a combination thereof, as will be
shown by typical designs. The output sequence of
the controller may be generated at a rate equal
to or at a multiple of the input sequence rate.
For linear systems, it is possible to program the
computer to cause the control system to have a
desirable over-all prototype response subject to
certain theoretical limitations. For computers
in which the output data rate is a multiple of
the input rate, faster response can be achieved.
The theory of sampled-data systems is applied to
the synthesis of the digital controller program
which will produce these desirable prototype response functions. The theoretical restrictions
which limit the choice of response functions are
pointed out. Examples of typical designs which
illustrate the theory are given.

The Program of Technical Sessions of the IRE
National Convention in New York, March 1957, contains a number of papers having some relation to
computers and data processors. Following are the
titles and abstracts of these papers, and notations of the part of the IRE Convention Record in
which they will be published. Copies of this
publication will be obtainable from The Institute
of Radio Engineers, Inc., One East 79 St., New
York 21, N.Y.
SESSION 6
Sponsored by th~ Professional Group on Communications Systems. To be published in Part 8
of the 1957 IRE Convention Record.
MULTIPLEX COMMUNICATIONS SYSTEMS
6.2

Time-Division Multiplex System with Addressed
Information Packages - R. Filipowsky and E.
Scherer, Westinghouse Electric Corp., Baltimore, Md.

9.4

A digital multiplex communication system is
described wherein the channels are combined in
asynchronous time division; there is no fixed and
predetermined sequence by which the individual
transmission interval is electronically "auctioned"
and is allotted to the highest bidder, i.e., to
the channel having instantaneously the most urgent
need for forwarding its message or a fraction
thereof. The receiver has no knowledge to which
channel the information within one transmission
interval should be directed, unless each interval
will carry its own address. One interval may contain one quantized sample only or a larger amount
of time compressed information, i.e., one information,package.

It has been widely accepted that many processes in the broad field of science essentially
can be classified as closed-loop systems. Many
of the problems encountered in these fields are
essentially statistical in nature and, as a result, have not been analyzed using the differential
equation approach; however ,_ au~ocorrelation and
spectral density approach have been used. ~t is
demonstrated that the differential equation'
approach can be used to determine the expected
behavior of many of the "statistical type systems."
SESSION 22
Sponsored by the Professional Group on Information Theory. To be published in Part 2 of the
1957 IRE Convention Record.

SESSION 9

INFORMATION THEORY--APPLICATIONS

Sponsored by the Professional Group on Automatic Control. To be published in Part 4 of the
1957 IRE Convention Record.

22.1 An Inductive Inference Machine - R. J.
Solomonoff, Technical Research Group, New
York, N.Y.

AUTOMATIC CONTROL--GENERAL
9.1

Solution of Statistical Problems by Automatic
Control Techniques - R. L. Cosgriff, Dept. of
Elec. Eng., Ohio State University, Columbus,
Ohio

A study has been made of a category of machines
that will perform inductive inferences. A simplified model will be described that only uses a few
of a more complete set of heuristic devices. Such
a machine is able to-~earn arithmetic operations
from a small set of examples.
By using all of the heuristic devices, it is

Digital Controllers for Feedback Systems J. R. Ragazzini, Columbia University, New
York, N.Y.

A digital controller for a feedback control
system is a computer which accepts a sequence of
numbers at its input and delivers a processed
- 30 -

IRE - Abstracts

expected that a machine will be able to learn to
perform complex tasks for which it was not specifically designed. 'Proving theorems, playing good
chess, and answering questionsin English appear to
be within ultimate machine capabilities.

the use of 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 2 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 inches and magnetic heads which provide up to 18 recording tracks.
The weight-type reel servo system, the highcurrent transistor solenoid driver, and the tape
control logic are also described. Several unique
features of this equipment have been employed in
this developmental equipment to achieve performance and enhance its reproducibility.

22.2 Multicase Binary Codes for Nonuniform Character Distributions - F. P. Brooks, Jr., International Business Machines Corp., Poughkeepsie, N.Y.
For economical transmission, variable-length
coding systems are theoretically best for nonuniform character distributions, but difficulties of
decoding and checking limit their practical usefulness. A useful class of variable length codes
are the multicase codes, in which the characters
(messages) to be represented are divided into two
or more cases, with shift characters signaling
changes in the case of succeeding characters.
Analysis of the properties of multicase coding
systems yields a method of generating and selecting economical variable-length codes that can be
readily encoded, decoded, and checked.

26.2 A Magnetic Pulse-Current Regulator - J. D.
Lawrence, Jr. and T. H. Bonn, Remington Rand
Univac, Philadelphia, Pa.

SESSION 23
Sponsored by the Professional Groups on Electron Devices and Broadcast and Television Receivers.
To be published in Part 3 of the 1957 IRE Convention Record.
TELEVISUAL SYSTEMS DEVICES
23.4 An Electrostatic Character-Writing Tube K. Schlesinger, B. Maggos, and A. F. Hogg,
Motorola, Inc., Chicago, Ill.
A 19-inch charactron tube has been developed,
which uses electrostatic fields throughout for
character selection and distribution, as well as
for image formation and intensification. Two
pairs of conjugate electrostatic yokes are used
for letter-selection, and a fifth unit for character-positioning at wide angles (70 degrees).
A long decelerating electrostatic lens is
used for image formation, permit~ing instantaneous
change of size between small and capital letters.
A new barrier-mask accelerator has been developed to increase beam velocity twelve-to-one
over the last inch of its path ahead of the screen.

This paper describes a magnetic current regulator using a square hysteresis loop core with two
windings. A direct current flowing in one winding
holds the core in a saturated condition and provides a current reference for regulation. The
core presents a low impedance to a current pulse
passing through the other winding until the pulse
mmf exceeds the dc mmf. When this happens, the
core moves into the unsaturated region of its
hysteresis loop and presents a high impedance to
the current pulse, thereby preventing further increase in this current ,pulse.
Design considerations are presented. The
precision of regulation is directly proportional
to the number of turns on the pulse winding and
inversely proportional to the magnetizing mmf required by the core in its unsaturated region.
Hence, precision of regulation is limited by the
amount of air core inductance that can be tolerated and the number of turns that it is physically
possible to place on a core. The use of a thinwall metal bobbin substantially improves the regulation.
26.3 Diodeless Magnetic Core Logical Circuits L. A. Russell, IBM Research Center, Poughkeepsie, N.Y.
Magnetic cores having rectangular hysteresis
loops have been shown to be useful as a key element in logical circuits for digital computers.
However, most of these circuits use diodes for
coupling, requiring windings of large numbers of
turns, or large cores. A special class of magnetic core logical circuits will be presented in
which cores are used for coupling. By avoiding
the use of diodes, these circuits offer major advantages in economy, reliability, and compactness
in medium speed applications. The basic technique that eliminates the need for diodes requires
that some of the cores have a switching threshold.
Specific examples of possible circuits and their
operating characteristics will be shown.

SESSION 26
Sponsored by the Professional Group on Electronic Computers. To be published in Part 4 of
1957 IRE Convention Record.
ELECTRONIC COMPUTERS I--DIGITAL COMPUTERS
26.1 An RCA High-Performance Tape Transport System - S. Baybick and R. E. Montijo, Radio
Corp. of America, Camden, N.J.
A high-performance, multichannel digital tape
transport has been developed to meet the needs of
the data processing industry in general. This is
tubeless equipment which provides very fast start
and stop times at very high- repetition rates through
- 31 -

26.4 Digital Computer Designs Circuit for Longest
Mean Time to Failure - J. Alman, P. L. Phipps
and D. L. Wilson, Remington Rand Univac, Div.
of Sperry Rand Corp., St. Paul, Minn.
This paper describes a system of circuit design utilizing a digital computer. The digital

COmputers and Automation

computer is programmed in such a manner so that
it can compute the circuit that w~uld give the
longest mean time for failure of the circuit.
This is accomplished by programming into the
computer the characteristics of life test of components. The computer computes the circuit many
times looking for the combination of circuit components that would still meet the output requirements and give the longest mean life to failure.
The main computer output then becomes just the
circuit which would give the longest mean time to
failure possible with the existing component characteristics which were programmed into the computer.
26.5 Considerations in the Design of Character
Recognition Systems - E. C. Greanias and
Y. M. Hill, International Business Machines
Corp., Endicott, N.Y.
The basic factors pertaining to the Character
Sensing problem are,discussed. These are range of
style, range of quality, and number of characters
and symbols to be recognized. Possible defini~ions of ch~racter quality and units for measuring
It are described. The information handling steps
of the recognition process are outlined as:
1) Conversion of marks on paper to electrical
signals.
2) Discrimination between signal and noise.
3) Reduction of filtered data.
4) Identification of characters and symbols
based on reduced data.
5) Validity checks.
The use of digital computers in the simulation of data reduction and recognition is described. A method for generating realistic test
specimens on a digital computer is outlined.
SESSION 28
Sponsored by the Professional Group on Nuclear Science. To be published in Part 9 of the
1957 IRE Convention Record.

SESSION 35
Sponsored by the Professional Group on Audio.
To be published in Part 7 of the 1957 IRE Convention Record.
SPEECH ANALYSIS AND AUDIO AMPLIFIERS
35.1 A Demonstration of the Representation of
Speech by Poles and Zeros - S. H. Chang and
R. Bach, Jr., Northeastern University, Boston,
Mass.
Aside from the phonemic description of speech
the most effective representation of speech is
supplied by the dynamic model. The latter concept
of excitation functions and system functions is
very appealing to communication scientists. Work
remains to be done to demonstrate how successful
this concept is in solving speech compression
problems. This paper demonstrates our progress
to date.
SESSION 40
Sponsored by the Professional Group on Production Techniques. To be published in Part 6 of the
1957 IRE Convention Record.
PRODUCTION TECHNIQUES
40.2 An Approach to Airborne Digital Computer
Equipment Construction - P. E. Boron and
E. N. King, Hughes Aircraft Co., Culver
City, Calif.
This paper is a discussion of one method of
building airborne digital equipment, making use
of the unitized etched wiring plug-in philosophyand utilizing an all-etched wiring harness to
make the large number of connections between plugin units. Points of emphasis are miniaturization,
reliability, small weight, accessibility, and
manufacturability of the equipment.

NUCLEAR INSTRUMENTATION
SESSION 41
28.3 O.l-Microsecond, 2000-Channel, Electrostatic
Storage System for Time-of-Flight Experiments
- J. Hahn, Dept. of Physics, Columbia University, New York, N.Y. \

Sponsored by the Professional Group on Electronic Computers. To be published in Part 4 of
the 1957 IRE Convention Record.
ELECTRONIC COMPUTERS II--SYMPOSIUM ON COMPUTERS
IN
SIMULATION, DATA REDUCTION, AND CONTROL

An instrument has been developed for measuring neutron time-of-flight which has 2000 discrete
time channels, each having a width of 0.1 microsecond. A temporary electrostatic storage is
used in conjunction with a magnetic drum memory.
Neutrons initiated by a cyclotron burst, which
occurs at approximately a 60-cps rate, are detected after traversing a fixed flight path, and
their times of arrival are stored in the electrostatic memory. Before the next burst occurs the
stored information is read into the drum, and the
storage tube is cleared. This paper describes the
electrostatic storage system and circuitry~ The
major system blocks which are discussed are: the
10-mc pulsed oscillator and clock pulse generator;
the "clock-stopper" or staticizer; the fast carry
flip-flop deflection circuit drivers; the current
adder stairstep deflection generators, and the
storage tube.

Chairman: R. D. Elbourn, National Bureau of
ards, Washington, D.C.

Stand~

A rapidly expanding class of computer applications that includes the simulation of complex
systems in real time, on-line reduction of experimental data, and automatic process control is introducing new problems that were not met in the
mathematical and business applications.
Problems of speed and control affect the design of the computer itself. For sufficient speed
it may have to be wholly or partly analog or, if
digital, it may have to be a special purpose rather
than general purpose machine. The control may have
to permit interruption for special tasks and then
return automatically to a former task.
- 32 -

IRE -

Abstracts

Communication between the computer and the
rest of the system may involve problems of conversion between analog and digital data, of multiplexing many data channels into one computer channel, of smoothing or interpolating sampled data,
or of preparing displays suitable for human beings.
The first two speakers will deal generally
with these problems, then three speakers will describe their solutions in-three specific applications:
1) A large combined analog-digital simulator.
2) The digital control of machine tools.
3) The reduction of wind-tunnel data.
After these talks there will be a round table
discussion of present and future solutions of these
problems.
SESSION 44
Sponsored by the Professional Group on Industrial Electronics. To be published in Part 6 of
the 1957 IRE Convention Record.
INDUSTRIAL ELECTRONICS

Rome Air Development Center, Griffiss Air
Force Base, N.Y.
As the demands of modern warfare have dictated increases in electronic equipment complexity, reliability has been correspondingly decreased, thus necessitating the institution of a
comprehensive program of reliability improvement
in the Department of Defense. This paper describes
the factors leading to the development of the program at the Rome Air Development Center and the
methods of its implementation on radar and other
ground electronic equipment developed by RADC.
The RADC Reliability Program--past, present,
and future--is discussed. The general areas
covered are: 1) Methods of educating design
engineers on reliability; 2) development, verification and use of a reliability prediction
technique; 3) development of automatic monitoring
equipment; 4) component and circuit reliability
improvement techniques, and 5) insertion of quantitative reliability requirements into equipment
specifications.
45.2 A Reliability Program - R. E. Kuehn, International Business Machines Corp., Owego, N.Y.

44.1 The Canadian Mail Handling System and the
Problem of Coding - M. M. Levy and A.
Barszczewski, Post Office Dept., Ottawa,
Ontario, Canada
The Canadian Post Office has under development an Electronic Automatic Mail Sorting System.
The efficient operation of the system is made
possible by use of a proper code and coding methods.
The proper names of towns, villages, and
streets are converted into a special code suitable for electronic handling.
The process of coding is performed mentally
by the operators according to a set of simple
rules. The addresses of letters are read by the
operators in special reading stations and mentally coded, the code is then marked on the envelopes by a speCial keyboard. Afterwards the remaining operations are performed completely automatically.
The prototype of the system is briefly described. The paper is dealing mainly with the
coding problem and human factors involved in it.
A clear account is given of various codes and
coding aspects. Description is given of minimization of errors. The realization of a working
system did depend to a large extent on suitable
coding and therefore its importance is stressed.
Although this paper is restricted to the
specific application only, nevertheless the methods used have large potential possibilities in
other fields where realization of electronic sorting and filing of data is essential.
SESSION 45
Sponsored by the Professional Group on Reliability and Quality Control. To be published in
Part 10 of the 1957 IRE Convention Record.

The introduction of complex electronic systems into general usage has made the organization of reliability engineering groups a necessity. This group can function best by reporting
to the manager of engineering as an independent
evaluation and service agency. The tasks of '
reliability engineering include the development
and application of reliability prediction techniques; the operation of a failure reporting and
analysis system; the selection, qualification,
and application of components; the environmental
and life testing of components, units, and systems,
and the evaluation of systems on the bench and in
the field for reliability, maintainability, accuracy, and operational suitability.
45.3 A Reliability Program for Rand D Projects E. F. Dertinger, American Bosch Arma Corp.,
Garden City, N.Y.
The proposed paper will describe a long-range
reliability program which has been inaugurated
during the initial design stage of developing and
producing an inertial guidance system for ballistic
missile application. This program represents, to
the writer's knowledge, the first all-out attempt
to "design-in" complete missile system reliability.
Considerable effort will be extended toward
presenting the procedures and techniques developed
for "comparative evaluation" of functionally-suitable component parts and "reliability qualification" of: 1) parts selected for design incorporation, 2) assemblies and major components, and 3)
complete systems. The philosophy of Arma management as regards reliability and the support given
to this program will be described.

RELIABILITY PROGRAMS

45.4 The Role of Quality Engineering in Procuring
and Producing Reliable Products - R. A.
Hulnick, International Business Machines
Corp., Kingston, N.Y.

45.1 Air Force Ground Electronic Equipment Reliability Improvement Program - J. J. Naresky,

To assure production of a reliable product,
quality engineering must formulate for the manu- 33 -

Computers and Automation

facturer a plan which encompasses a design of acceptable reliability, adequate controls over procurement and processing, and techniques for continuous performance evaluation for the purpose of
product improvement.

The design now permits recording of standard
5, 6, 7, or 8-hole code patterns to be reliably
performed at controlled rates up to 240 codes per
second. The perforator executes a basic recording cycle of one code in approx~mately 4 milliseconds. Since a recording cycle can only be
executed on demand associated electronic control
circuits will permit recording of random or asynchron"ous data at variable rates up to 240/ second.

SESSION 46
Sponsored by the Professional Group on Telemetry and Remote Control. To be published in Part
5 of the 1957 IRE Convention Record.

46.4 A High-Speed Binary-to-Binary-Decimal Translator - C. A. Campbell, Radiation, Inc.,
Melbourne, Fla.

SYMPOSIUM--DIGITAL TECHNIQUES FOR PROBLEMS
IN
TELEMETERING AND REMOTE CONTROL

A high-speed translator has been developed
which accomplishes binary-to-binary-decimal translation yet requires only about three vacuum tubes
per bit.
By suitably arranging the feedback and trigger logic circuits the binary decades accept
counts with weights of 1, 2, 4, .6, or 8. With
minor circuit changes the counter will operate
with any integral weight from 1 through 9.
Arranging these decades in cascade, inputs
with weights of 16, 32, 64, etc., are accepted
and totaled to decimal readouts, which may be
in either binary or decimal presentation.
The present decade requires about five microseconds per binary digit for computation; however,
the state of the art can allow an increase of at
least tenfold in this speed.

Chairman: C. H. Hoeppner, Jr., Radiation, Inc.,
Melbourne, Fla.
46.1 A High-Speed Digital Data-Handling System G. F. Anderson, Radiation, Inc., Melbourne,
Fla.
This paper describes a high-speed digital
data-handling system which provides digitizing,
editing, and processing of pwm and fm/fm data.
This system digitizes a maximum of 22 channels of
analog data at a channel-sampling rate of 22,000
data points per second. An over-all accuracy of
one part in 1024 is provided, utilizing a 10-bit
binary straight code. Simultaneous sampling of
the data to an accuracy of 33 microseconds is
provided. Editing is provided by the Quick-Look
analog recording system. The processing system
generates an output code compatible with the
IBM 650 computer; however, with slight alterations, the output format can be made compatible
with other types of digital computers.

46.5 Simplicity for Reliable Low-Cost Operation
in a Digital Data-Processing System - J. W.
Prast, Bell Aircraft Corp., Buffalo, N.Y.
A simple and straightforward digital dataprocessing system for telemetering purposes was
developed in 1954 and has been in operation since
that time for processing special precision information during test flights of guided missiles.
Experience gained in operational use will be
reported with emphasis on reliability and cost
aspects.
The system features 0.1 per cent accuracy,
inherent reliability, and low cost on the advantageous side with limitations in versatility
as a disadvantageous feature; it was originally
designed for processing fm-type information
essentially not changing over a sampling interval of approximately 50 milliseconds. Improvements and additional applications for processing
information in analog-voltage form and fully
automatic operation will be discussed.

46.2 Magnetic Tape Playback and Digital Conversion
of Telemetered Flight Data for Entry into
Digital Computers - G. C. Dannals, Radiation,
Inc., Melbourne, Fla.
Paper describes an automatic digital conversion and data processing system designed and built
by Radiation, Inc. for the AC Spark Plug Division
of General Motors Corp. Decommutated and separated pdm and fm/fm data recovered during magnetic
tape playback of telemetry records is electronically sampled and converted to ten-pit binary code
and further processed to suitable form for entry
into high-speed digital computers. Data editing,
intermediate recording, and processing control
features are incorporated. Other salient features
as regards system accuracy, high-speed-handling,
and operational flexibility are discussea.

SESSION 49
Sponsored by the Professional Group on Electronic Computers. To be published in Part 4 of
the 1957 IRE Convention Record.

46.3 Design Considerations for Super Speed Perforated Tape Digital Recording - J. Bellinger,
J. T. MacNeill, and C. F. West, Soroban Engineering Co., Melbourne, Fla.

ELECTRONIC COMPUTERS lII--MAINLY ANALOG
49.1 Computation with Pulse Analogs - N. Rubenfeld,
W. L. Maxson Corp., New York, N.Y.

Until recently, techniques for recording
digital data on perforated tape have been perfected to meet the requirements of the communication industry. The recorder described in this
paper probably represents the first tape perforator designed for instrumentation as well as digit?l computer output data recording applications.

A special purpose
been built to handle a
analog frequencies and
novel techniques. The
- 34 -

semidigital computer has
computation involving two
an analog voltage using
equation to be solved is

IRE - Abstracts

puter circuits. Illustrative examples of steepest
descent paths and root loci obtained on the computer are included.

X fb

fa
The computation is done in two basic stepsj
the first step is computing a repetition rate
fr = Elfa and the second step is multiplying two
frequency analogs, fr X-fD' The entire computer
is built using basic blocks, such as, flip-flops,
gates, blocking-oscillators, and is completely
transistorized. Because of the flexibility of
the basic concepts, the ,dynamic range of the computer is virtually unlimited.

49.4 Magnetically Controlled Counters - E. A.
Sands, Armonk, N.Y.

49.2 A Cyclic Digital-to-Analog Decoder - G. H.
Myers, Rome Air Development Center, Griffiss
Air Force Base, N.Y.
This paper describes a counting decoder which
converts a binary number into a duration-modulated
pulse many times in a computing cycle without sacrificing the economy of an ordinary, basic decoder.
The device, which is part of the TRADIC computer,
uses only transistors and achieves its features
by using the "cyclic" nature of binary numbers.
That is, if a binary number is placed in a register and a fixed amount is continually subtracted
from it, the original number will reappear after
a constant time interval. Decoding the same number many times has the effect of giving the decoder a considerably greater bandwidth, which is
an important consideration in control or servomechanism problems. The decoder also has certain
other features which make it desirable in control
applications.
49.3 An Automatic Analog Computer Method for
Solving Polynomials and Finding Root Loci L. Levine and H. F. Meissinger, Hughes Aircraft Co., Culver City, Calif.
Various analog computer techniques are available which provide the means for rapid solution of
polynomials with an accuracy suitable for engineering purposes. Methods heretofore described
in the computer literature, however, have the
shortcoming of requiring a step-by-step procedure
for finding the roots.
In this paper a new analog technique is discussed by which the roots are determined automatically. This is an application of the "method
of steepest descent". A suitably chosen function
W is minimized in the computer by a continuous
adjustment of the coordinates of a point in the
complex z plane until a stable equilibrium is
reached. This corresponds to a root of the polynomial. In a polynomial of nth degree n such
equilibrium points can be found. The coordinate
adjustment follows the gradient of the function
Wand therefore leads to the minimum at the
fastest possible rate.
The computer may be operated in two modes:
1) searching for individual roots from points
arbitrarily chosen in the complex z plane, 2)
tracking the roots while coefficients of the
polynomial are being varied. The latter mode is
ideally suited for plotting root loci in the complex frequency plane.
This paper contains a discussion of mathematical and practical aspects of the steepest descent method and gives several alternative com-

A magnetically controlled counter will be
described in which the count determining circuit
is a pair of magnetic cores. A simple theoretical analysis will be made using the principles
of equivalent core impedance. Some practical
circuits will be shown, and deviations from predicted behavior will be discussed. Methods of
designing units to produce reliable counts from
scale 2 to scale 16 will be indicated. Means of
presetting will be pointed out, and practical
limitations on the use of the device will be explained.
49.5 Systematic Tracing of Discrepancies in Analog Computers - M. Plotkin, Naval Air Development Center, Johnsville, Pa., and
E. Grosswald, University of Pennsylvania,
Philadelphia, Pa.
Large analog computers are for the most part
used in control, or closed loop, problems. In
problems of this sort an error in one location,
whether due to incorrect plugging or to a faulty
component, causes errors throughout the machine.
Should the errors result in observed discrepancies, by comparison of the machine output with
a reference solution either obtained independently or produced earlier by the same computer, it
is sometimes difficult in closed loop problems to
find the cause. This paper proposes a method for
locating systematically the source of such discrepancies.
SESSION 52
Sponsored by the Professional Group on Reliability and Quality Controlr To be published in
Part 10 of the 1957 IRE Convention Record.
ANALYSIS AND TECHNIQUES FOR IMPROVED RELIABILITY
52.4 Reliability Prediction Technique for Use in
Design of Complex Systems - H. E. Blanton,
Hycon Eastern, Inc., Cambridge, Mass.
A relatively simple reliability-prediction
technique has been developed which can be instituted during the early stages of the design of
complex systems. Through the use of reliability
diagrams (obtained by modifying engineering block
diagrams) and basic rules from probability theory,
reliability formulas for proposed designs are
derived. Effects of secondary failures and variations in requirements for successful performance
are included. By evaluating the reliability
formulas using the best available component-reliability estimates, alternative designs can be compared and the need for redundancy or component
improvement is established. An example based on
an airborne telemetering system illustrates the
technique.

- 35 -

N'EW PATENTS

RAY:\10ND R. SKOLNICK
Reg. Patent Agent
Ford Inst. Co. Div. of Sperry nand Corp.
Long Island City 1, New York

The following is a compilation of patents pertaining to computers and associated equipment from
the Official Gazette of the United States Patent
Office, dates of issue as indicated. Each entry
consists of: Patent number / inventor(s) / assignee / invention.

january 22, 1957 (contVd from Mar. issue):
2,778,624 / Louis D. Statham, Beverly Hills,
Calif. / Statham Laboratories, Inc., Los
Angeles, Calif. / An angular Accelerometer.
January 29, 1957: 2,779,538 / William M.
Shanhouse, Roslyn Heights, N.Y. / - / A
navigational relative motion computer.
2,779,539 / Sidney Darlington, Passaic Township, Morris County, N.J. / Bell Telephone
Laboratories, Inc., New York, N.Y. / A
multiple code wheel analogue - digital
translator.
2,779,655 / Waldo H. Kliever, Minneapolis,
Minn. / Minneapolis-Honeywell Regulator
Co., Minneapolis, Minn. / A graphical to
digital indication converter.
2,779,871 / Omar L. Patterson, Media, Pa. /
Sun Oil Co., Philadelphia, Pa. / A differentiating circuit.
2,779,872 / Omar L. Patterson, Media, Pa. /
Sun Oil Co., Philadelphia, Pa. / An integrating circuit.
Feb. 5, 1957: 2,780,408 / Loring P. Crosman,
Wilton, and Francis B. Hannon, Milford,
Conn. / Sperry Rand Corp., Del. / An electronic accumulator for adding and subtracting received digit values.
2,780,409 / George A. Haidenbergh, St. Paul,
Minn. / U.S.A. I A binary accumulator circuit of the parallel type.
2,780,410 / Rolland N. Bued, Clifton, N.J.,
Willard B. Groth, Bronx, N.Y., Gordon C.
Irwin, Fair Haven, and Lindley A. Kille,
Morristown, N.J., and George Riggs, Port
Washington, N.Y. / Bell Telephone Laboratories, Inc., New York, N.Y. / A tape to - card convecter circuit.
2,780,455 / Howard F. Devaney, Albuquerque,
N. Mex. / U.S.A. / A combination of acceleration sensing and integrating means.
2,780,672 / Johannes Anton Greefker and Piet
van Tilbuy, Eindhoven, Netherlands / Hartford National Bank and Trust Co., Hartford,
Conn. / A device for separating synchronizing
pulses and signal pulses with pulse-code
modulation. .

2,780,673 / John B. Singel, Catonsvflle, Md. /
Westinghouse Electric Corp., East Pittsburgh,
Pa. / A phasing system.
2,780,725 / Norman L. Johanson, Seattle,
Wash. / Boeing Airplane Co., Seattle, Wash. /
A modulator - demodulator limiter transistor
circuit.
2,780,760 / Warren E. Dion, Bristol, Conn. /
Bristol Co., Waterbury, Conn. / A servo system with quick - action response.
Feb. 12, 1957: 2,781,168 / Ivan A. Greenwood,
Jr., Stamford, Conn. / General Precision
Laboratory Inc., N.Y. / A great circle computer.
2,781,169 / John F. Donan, Los Angeles, and
Lawrence D. Hindell, Gardena, Calif. /
Northrop Aircraft, Inc., Hawthorne, Calif. /
Means for the electronic addition of vectors
at substantially any angle.
2,781,170 / Donald F. Walker, Hollinwood, Eng. /
- / An electrical computing instrument for
deriving the value of an unknown quantity
of a problem to be solved from three known
quantities, where the latter can be represented as elements of a triangle.
2,781,447 / Burton R. Lester, Camillus, N.Y. /
General Electric Co., N.Y. / A binary digital
computing and counting apparatus.
2,781,482 / James L. Montgomery, Indianapolis,
Ind. / U.S.A. / A balanced bridge type servo
system.
2,781,504 / Michele Canepa, South Norwalk,
Conn. / Olivetti Corp. of America, New York,
N.Y. / A binary system capable of providing
indications of the presence of a signal or
of its binary complement.
Feb. 19, 1957: 2,781,967 / Rolf E. Spencer,
Ealing, London, and Richard H. Booth,
Beaconsfield, Eng. / Electric and Musicol
Ind. Lim., Hayes, Eng. / An apparatus for
evaluating the instantaneous value of cos
a x where a is a constant and x is an independent variable.
2,781,968 / Pierce J. C. Chenus, Paris, Fr. /
Compagnie des Machines Bull (Socie't~ Anonyme),
Paris, Fr. / An addition and subtraction
operating device for electric calculating
machines operating in the binary system.
2,781,969 / Alexander Somerville, Birmingham,
Ala. / - / An apparatus for determining the
product of a pair of variable quantities
expressed in input voltages proportional
to the quantities.
2,781,970 / Sidney Kaufman, Houston, Tex. /
Shell Development Co., Emeryville, Calif. /
An electronic analog computer.
2,782,347 / John A. Herbst, Montville, N.J. /
Bogne Electric Manufacturing Co., Patterson,
N.J. / A fail - safe servo system.
2,782,408 / Walter W. Fisher, Pacoima, and
Carl E. Sohlgren, San Fernando, Calif. /
Bendix Aviation Corp., North Hollywood,
Calif. / An apparatus for translating reflected binary code multidigit signals into
corresponding impedance values between
first and second main terminals.
(cont'd on page 37)

- 36 -

Forum
MAGAZINES ON "COMPUTOLOGY"
Samuel J. M. England
Columbus, Ohio
I have recently joined the staff of the
Systems Engineering Division at -----, where
I am engaged in making a study of the computer
field. I have not previously acquainted myself with your publication, but now I have sat
myself down and digested each (available) issue
from the beginning tp the present.
I am sufficiently impressed by the "tone"
of "Computers & Automation" to be moved to
write to you. You will be the best judge of my
bias; -my feelings are as follows:
I have for the past two years read most
of the issues of "-----", "-----", "-----,,
and "-----". I subscribed to one of them,
"-----", and receive free copies of another,
"-----,,
My feeling about these journals is
that for the most part they are rather poorly
disguised efforts to get advertizing into one's
hands. Though this statement is too harsh to
be entirely true, the fact remains that they
serve some other purpose than that which meets
with my approval.
In contrast, your publication has improved
rather than deteriorat~d since Jan. '54 (the
first copy that I have). The thing that I
commend is a certain evidence of courage on
the part of your staff for such articles as
may appear to some to be a non sequitur--for
example, the article No. 2 under Forum in the
Jan. 1954 issue, page 14, entitled "Shakuntala
Devi, Mathematical Prodigy". In fact, when I
ran across your editorial comment in April
1954, 'which I will quote, I was by then assured
that you indeed have this intention:
"In the pages of this magazine we shall
do our best to promote controversy, honorable
controversy, which tries to make sure that
each side of a question is expressed fairly~
without calling names, attacking reputations,
or hugging orthodoxy."

I detect a feature which should be exploited in order to further the influence of
your journal. I will cite two examples: In
your January 1957 issue there is an acceptably quantitative article by Walter F. Bauer
entitled "Modern Large Scale Computer System
Design". This article and its kind must be
made available to budding computologists to
act as a bridle to their untried speculation
about computers. In comparison to this, I
was impressed also by what might be called
a more literary article which appears in the
October 1956 issue, page 37, entitled "The
Pure Word of St. Euphorus", by J. W. Granholm.
In this same vein but of a more sober or
scholarly nature is the article in the most
recent (March and April i57) issues dealing
with the History of Robots and a bibliography.
I am not entirely certain that you are
directed by anything but expediency because
of the lack of qualified manuscripts; but I
do not believe this to be so, and therefore
would congratulate you for the face that your
publication chooses to display.
I hope that your journal will be able to
continue carrying the spirit that is to be
identified with such people as Norbert Wiener,
John Von Neumann, Vannevar Bush, M. Turing,
and others, who would prefer the "human use
of human beings".
- END-

* --------- * ----------*
NEW PATENTS

(cont'd from page 36)

Feb. 26. 1957: 2,782,985 I Edward L. Vibbard,
La Jolla, Calif. I Bell Telephone Laboratories, Inc., New York, N.Y. I A tape control
arrangement for a computer.
2,783,421 I Karl W. Hering, Ridgefield, Conn.1
The Perkin-Elmer Corp., Norwalk, Conn. / A
compensated velocity servo-loop system.
2,783,422 / Marcel Fouassin, Liege, Belgium /
- / A preset servo system.
- END -

May you prosper in this design, though
the path is fraught with difficulty.
Another policy which I detect, which in
the long run will prove invaluable, is that of
talking to the audience. There is need in
technology and science for the Greek Dramatic
mechanism of the chorus. Do not suppress the
part of the chorus, for the audience need to
feel that they have a representative to mediate for them in the drama.
- 37 -

Forum

THE UNITED STATES MAILS

FIELD FOn AUTOMATIC

PROCESSING OF INFORMATION
I.

Most mail sorting in this country is now
done by hand, but the article suggests that
much of it could be mechanized. It states
that Holland began experimenting with letter
sorting machines in 1928, and the Dutch Post
Office now has 11 such machines each sorting
50 letters per minute to 300 or more destinations as opposed to 25 per minute to 50 destinations by hand.

From The Reader's Digest,
Pleasantville, N.Y.
A.

Release of April 8

The Post Office system "needs mechanizing and st:eamlining all the way through," a
Reader's DIgest article, titled "Our Horse and
Buggy Mails", will say in the May issue of the
magazine.
This will not appear until April 23. But
the publication today released a summary of
the 5,000-word, article by Wolfgang Langewiesche
when the magazine learned that someone in Washington without authority from Reader's Digest
had made the article available to certain members of Congress prior to the normal publication date.
"The present administration has begun
some mechanization but we are years behind
Europe," says the article which concludes with
the appeal: "Let's create a modern mail system."

New buildings in new locations--on the
outskirts, near airports--are needed, with
midtown buildings concentrating on local mail
moving only a few blocks between sender and
addressee. Helicopters could lift whole
trailers full of mail between sorting plants
and downtown locations.
Whether mechanized or not, the article
says, a post office building should be "a
sort of factory." It should be tall, so the
mail can flow down from floor to floor, or
else, it should be flat with a floor plan in
which conveyor belts can fan out like railroad tracks in a freight yard.
Instead, the article points out, we have
Greek temple, or a best a First National
bank" and many post offices date from the
last century. Many were built during the
depression "when the idea was to make more
work" and designed "mostly to express the
majesty of the federa,l government."
~',a

According to the article, which discusses
Post Office methods, buildings and policies
~he servi~e "uses the same methods of gathe~
lng, sortIng and delivering the mail that it
did 100 years ago" and mail is now "slower
than it was before World War II."

The article ends by stating that "in
the 20th century even this wealthy country
can't subsidize 19th-century mail systems
indefinitely."

The Post Office'is one of the country's
top 12 enterprises in volume of business and
t~ird in number of employees. But it st~adily
vIolates the business principle of "stay ahead
of the times--or die," the article states.

"The Post Office needs money for research
and development ••• It's asking for four million
dol~ars this year.
That's chicken feed! EspecIally ~hen, for lack of research, it spends
--beyon~ Its income--half that much a day!
ProportIonately, the Telephone Company spends
18 times as much on research!"

The first part of the Article:
"Our Horse and Buggy Mails"

By annual business done--three billion
dollars--the U.S. Post Office ranks among
A~erica's top dozen enterprises, among such
gIants as Ford, General Electric, U.S. Steel.
By number of employes--half a million--it
ranks third, right after the Telephone Company and General Motors. It handles 60 percent of the world's mail. It has 40,000
branch offices. Every day it carries one
piece of mail for every man, woman and child
in the country.

W~ges take 75 percent of postal costs.
Mec~anlzation would cut the costs, says the

artIcle, and enable the Post Office to stay
on top of the mail flood. But instead of
machines, men are used in "primitive proced~res" that. waste ti~e, effort and money. '
In other Industries, the bigger the volume,
~h; lower the cost," the article says.
"But
It s a technological law: 'you're supposed to
work with machines--bigger, faster, more efficient machines."

This makes the Post Office one of the
world's biggest industrial operations. But
between real business and Post Office business there is one big difference:
- 38 -

u. s.

~ails

Every American business knows that it
must stay ahead of the times--or die. This
is the principle that has broug~t us from
the hand workshop to the automatic factory.
Today we are in the age of the jet, of lightning
communication, but the Post Office is not with
us. It still plods along, horse and buggy
fashion. It uses the same methods of gathering,
sorting and delivering the mail that it did
100 years ago. The result is something close
to chaos.
The mail is slower than it was before
World War II. A letter often takes 48 hours
to travel 100 miles. One letter, mailed Tuesday morning, was not received seven miles away
till Thursday afternoon. Circulars in New York
City go undelivered for as long as ten days.
A letter mailed in New York Wednesday noon
took until Monday afternoon for delivery In
Chicago. A letter to Chicago mailed from
Glencoe, 18 miles distant, took three days.
A few years ago residential mail deliveries
were cut to one a day. So now, if a letter
misses connections by a minute, it misses by
24 hours.
The Post Office is floundering in a sea
of mail that gets deeper every year. Postal
volume has risen steadily in the last 20 years.
(It has doubled in New York City.) This 'is a
flood situation; it calls for swift and basic
action. The Post Office has tried to keep
pace, but despite its efforts it keeps falling
behind. And the price we pay is staggering.
The Post Office is spending two million dollars a day more than it receives in revenue.
Even old-fashioned, slow-going Europe has
kept pace better than we have. In London, to
speed the handling of the mails, the Post
Office has its own subway, from one end of the
city to the other. A housewife can mail her
grocery order in the morning and get stuff delivered the same day. A week-end guest's
bread-and-butter letter arrives so promptly
that an American exclaimed: "He must have
mailed it before he came:" A letter mailed in
Paris before 6 p.m. is delivered the next day
anywhere in France. When it is necessary to
catch the early local trains departing.from
provincial centers special mail planes fly it
at no extra charge. In Munich you can get a
special-delivery letter to anybody within an
hour.
Our Post Office performance also contrasts
strangely with private enterprise here at home.
United Parcel Service--the firm that delivers
packages for retail stores--also has a service
for wholesalers. It sometimes charges less
than the Post Office would to take the same
package to the ~ame address. Further, it picks
up at the sender's. It promises a delivery
date (the Post Office doesn't). It makes

three attempts at delivery (the Post Office
makes only one). It throws in $100 insurance,
free. And it makes a profit, and it pays
taxes! ~n parcel post the Po~Office just
about breaks even.) You might say--yes, but
the United Parcel picked the juiciest part of
the parcel business. Not so. They have to
fight urban congestion. Besides, they have
just started to serve two entire states,
Illinois and Massachusetts, from metropolis
to the quietest back-country farm. It recently started a coast-to-coast air parcel
service--at half the rate of air parcel post!
Also, at rates comparable to ground parcel
post, it airlifts parcels coast-to-coast on
a space-available basis in four days, beating the U.S. Mails by a week!
Why is the Post Office so far behind
the times? The reasons are many ••••
II.

From Arthur E. Summerfield,
The Postmaster General,
Washington, D.C.
A.

Form letter of April 18

Dear Postal Patron:
The President has asked me to reply to
your telegram concerning our postal problem.
The attached press statement announcing
the resumption of service, with the accompanying fact sheet, will, I trust, answ~r
some of your questions.
Thank you very much for your interest in
this all-important subject.
B.

Excerpts from Release of April 16

I am happy to announce the resumption of
normal mail service and gratified to have the
overwhelming affirmative vote of the Congress
giving the Department funds for this purpose.
The legislation passed yesterday by the
House and this noon by the Senate gives the
Post Office Department $41 million for its
operations through June 30.
Within 24 hours mail service will be back
on nearly the same basis it was prior to the
issuance of our Order last Friday.
The need for supplemental appropriations
is common and often unavoidable in the Post
Office Department. In the past 20 years
they were needed in all except 3 years during
which the present postal administration returned $235 million of unspent money to the
(cont'd on page 43)

- 39 -

This form was used for our March survey,.

SURVEY-ESTIMATE
OF

THE

COMPU,TER

See page 8.

Can you estimate(roughly and approximately)
about how much your organization is likely to spend
on products and services in the computer field
2.

M,ARKET

The computer field contains many hard-toanswer questions, which can be partly answered
through surveys. The purpose of this first survey
being made by "Computers and Automation" is to
form an estimate of the size of the market for computers, data processors, and related equipment.
Following is the inquiry form for this survey. The
response of any person who considers himself- in the computer field IS welcome, and will be
much appreciated. The form may be torn out of the
magazine, or may be copied on any piece of paper.
We hope that the results when published will be of
use to all our readers.

- in the next twelve months?
between $
and $_ _ __
- in the next five years?
between $_ _ _ _ _ and $_ _ __
3. What perplexing questions or subjects would
you like us to inquire about in our surveys?

_ _ _ _ _ _ _ _ _(attach paper if needed)
4. Any remarks? _ _ _ _ _ _ _ _ _ _ _ __

ESTIMATE OF THE MARKET
FOR COMPUTERS AND DATA PROCESSORS
Questions
1. What kinds of computer products and services
does your organization buy or rent (or is considering buying or renting)?

And for statistical purposes: Your department?

Your chief job responsibilities? _ _ _ _ _ __
-

Computers
Yes
No
automatic digital computers? ( ) ( )
automatic analog computers?
(
) (
)
simulators?
)
(
) (
other data processing mach(
) (
)
ines?
such as: _ _ _ _ _ _ _ _ _ _ _ _ _ __

Do your recommendations affect purchases?

Your organization I s main products ? _ _ _ _ __

Components
No
Yes
delay lines?
(
) (
)
magnetic tape devices?
(
) (
)
transistors?
(
) (
)
other components?
(
) (
)
such as: _ _ _ _ _ _ _ _ _ _ _ _ _ __

No. of employees? _ _ _ _ _ _ _ _ _ _ __
Filled in by: Name _ _ _ _ _ _ _ _ _ __
Title _ _ _ _ _ _ _ _ _ _ _ _Date_ _ _ __
Organization,_ _ _ _ _ _ _ _ _ _ _ _ _ __
Address__________________

Services
Yes
No
- computing services?
(
) (
)
- consulting services?
(
) (
)
- other services?
(
) (
)
such as: _ _ _ _ _ _ _ _ _ _ _ _ __

When completed to the extent you conveniently
can, please return this sur ve y for m to
E. C. Berkeley, Editor, Computers and Automation, 815 Washington St., Newtonville 60, Mass.

- 40 -

NEW PRODUcrS AND IDEAS

(cont'd from page 19)

and has become known as a swift source of electronic computer components (within the lines
that it covers). Its speed derives from its original idea, the structure of its· organization,
the system of inventory (which does not use
automatic data processing machinery), and its
experience.
For example, a large Chicago company called
one morning for some special connectors needed
at once to complete work on an experimental computer for the government's guided missile program. Customarily each day's orders are processed by 3:00 p.m., but since this was an emergency, the complete order, filled out of stock
on hand, was delivered to LaGuardia Airport in
time for the 11:00 a.m. plane to Chicago. The
flight number was telegraphed via Western Union
Desk Fax, and the'purchaser picked up the material at the Chicago Air Terminal about 2:00 p.m. Chicago time. Before the day was over,
the connectors were being installed, and the
delay was eliminated.
Again, a short time ago, a call from
Bourns Laboratories in California came in, to
verify the supply of a particular item, a
trimpot.
It turned out that General Precision Laboratories of Pleasantville, N.Y.,
was holding another long distance line to the
Bourns factory, while Bourns phoned Schweber.
Schweber, with an ample stock of the particular trimpot needed, was able to send the
whole order to Pleasantville immediately.
Bourns Laboratories said, 'We feel that
Schweber Electronics is an outstanding example of a new type of distributor in the electronics field. These distributors handle
relatively few lines and carry very complete
stocks. This allows each person in their organization to spend a greater amount of time
in acquiring and disseminating knowledge of
each particular line. We feel that the company has established a national reputation for
always having any particular item in the lines
that they handle."
The company in fact acts as a warehouse
for its customers, reducing their investment
in inventory, storage space, and maintenance,
a nd their risk of obsolescence of components due to rapid change in design.
,
When tantalum qapacitors were first developed, only a few types had been manufactured: But it appeared to the company that
the aIrcraft and guided missile field would
soon require a new high temperature capacitor.
So they placed their own order for an additional 100 types in the high temperature range
to meet this forecasted need. The investment
was made in a new line with no known demand
and no real assurance that demand would de- 41 -

velop. , Delivery at the time of the company's
order was five months.
Tantalum capacitors are now an integral
part of every guided missile. For a time,
the company was the only source of supply that
could, from stock, deliver high temperature
tantalum capacitors when suddenly demanded by
industry. Accordingly, many prototypes for
important defense projects were completed without five months of delay.
Ninety-eight percent of the company's
business is conducted by telephone. There is
no switchboard; calls go directly to any of the
eleven sales specialists, who are trained to
handle at once any and all customer requirements. This sales staff know their ten lines
(made up of ten thousand different items) ,
and are able to make recommendations to engineers as to the correct component for their
need.
The company maintains an apparently crude
but effective and quick inventory control. With
a glance at the card record, the stock on hand
of any particular item is revealed, also how
many are on order, when material is due in,
how many units are committed by previous orders,
and which customers use this item. Therefore,
a customer placing an order, can be told at
once how soon he can expect delivery in order
that his engineers may schedule the production
accordingly. There are no "runners" or "stock
men". Each salesman posts the running record
on the inventory card. Since his relationship
with his customers depends on his accuracy,
this record stays accurate at all times.
Two large punch card manufacturers were
·both asked to improve the inventory control,
and were requested to investigate. After an- .
alysis, they said they could not. The reason
basically is the up-to-date-ness of the entries
made by the actual man who deals directly with
the customers -- with no intervening key punch
operators, supervisors of key punch operators,
etc.
The volume of business at the company has
doubled each year for the past three years. To
better meet the growing need of the computer
and missile industry, the company plans to build
a new plant and also establish a west coast
branch.

nr..LIAIHLll r

(cont'd from page 23)

ion will we be adequately assured that the
possibility of total failure has been eliminated and that operations can be safely entrusted to the system.

have found it more economical to run the computer until it breaks down before calling for
the engineers. It seems obvious, however,
that this practice could not be tolerated in
our system.

- END-

*---------Forum
*---------- *

The mere availability of "backup" or reserve equipment does not provide insurance.
Since the computer in our system f~nctions as
the nerve center of a complex communications
network, it will generally be impossible to
transfer operations to another machine at a
distant location without severely disrupting
the information flow. However, commercial
teletype lines and neighboring printer units
can be utilized as temporary replacements for
certain system components.

ANALOGUE COMPU'IERS IN EUROPE
P. A. R. Wright
Short Brothers & Harland Ltd.
London, England
We were interested to read in the February edition of "Computers and Automation", the
article by Everett S. ~alhoun on "New Computer
Developments Around !-he World".

A fundamental requirement in our system
design is the provision of means for human
intervention. Intermittent failures in system components are to be expected; their
functions must be performed during the interval necessary to correct and replace the mechanical faults. At present, an intelligent
human being is the surest substitute and the
safest means of providing against the varying
and complex difficulties that can arise. Accordingly,it is necessary to provide the mechanical system with a capable and trained
operating staff. This staff must be thoroughly aware of precisely what the system is
accomplishing so that it can immediately
introduce the correct remedial steps in case
of failure. We have found it wise to run
periodic "alerts" in which a variety of failures are simulated and the proper countermeasures are promptly initiated.

Whilst his article was extremely well informed, we were surprised to note that it was
confined almost exclusively to digital computing equipment with little reference to progress
in the analogue computing field.

Application of the principles of automatic control to the operation of business
organizations presents an exciting opportunity. Commercial data processing systems designed to utilize these techniques will assume responsibility for the direction and
control of all routine daily functions. The
use of comprehensive communications networks
will enable a central data processing unit to
exercise effective supervision over operations
of the entire organization. The incorporation
of feedback loops in the system will provide
a means of notifying the data processing center of the results of earlier activities.
Management reports will be abstracted from
the flow of daily information, thus providing
the executive with an accurate analysis of
the current situation.
As such real time business systems as-'
sume more overall responsibility and control,
an even greater stress must be placed on the
reliability of their operation. For failure
in these systems implies not just delay and
annoyance but almost complete paralysis of
activity. It is imperative that extreme.
care be exercised in the design and operation
of on-line systems so that all component failures are provided against. Only in this f?sh-

- 42 -

The use of Analogue Computers in Europe
is increasing remarkably, and is a subject
which we think you should mention in future
editions of your Journal. Such Analogue Computers are in use in all aircraft companies
in this country and in several of the aircraft
firms in France. We ourselves are using eight
computers of the type described in the enclosed
brochure, and twenty-seven computers of this
type are also employed by other aircraft firms
in this country, by Fiat Mirafiori in Turin,
Italy, by the National Luchtvaartlaboratorium,
Amsterdam, by the Technische Hochschule, ~raun
schweig, Germany, and by Stenhardt Ingeniorsfirma AB in Stockholm.
In Mr. Calhoun's article there is mention
of Darmstadt Technische Hochschule which is
regarded as the computing centre in Germany.
Although this University has been engaged
primarily on digital computing techniques,
there has been some considerable research there
by Dipl.-Ing. W. Dehn on multiplying devices
and repeti tively drift-corrected
0 pe ra tional amplifiers for Analogue Computers. Most
Analogue Computer work in Germany, however, is
currently undertaken by the Technische Hochschule Braunschweig, under the direction of
Dr. Horst Herrmann of the Institut fUr Angewandte Mathematik.
As regards computer production in Germany,
it is no longer true to say that the only work
is in "a barn in Neukirchen-Hiinfeld".
Apart
from our computer which is now in use in Braunschweig,there is under development by Telefunken,
in Ulm, a general purpose Analogue Computer;
Analogue Computers have also been manufactured
by the German firm of Schoppe und Faeser and
by the American subsidiary Company of Beckman
Instruments, G.m.b.H. in Munchen.
We hope the above brief comments will be
of some interest to your readers.

U. S. MAILS
<

(cont'd from page 39)

Treasury (in 1953, 1954 and 1955) out of funds
appropriated to the Post Office Department by
the Congress.

["BURROUGHS
RESEARCH' CENTER
NEED's
'Good
ENGINEERS
L,
'
,
,
"
*

THAT

Let me say that the letters the public has
sent us during the past 10 days are also appreciated. By an overwhelming majority, they have
endorsed our actions and many of them have indicated their willingness to pay more postage
so that they might get better mail service •

CERTAIN
MAN

How much does the Post Office Department
s pend in a year?

A: About $3 billion, 250 million. Of
this amount, 78.4 percent goes for postal employees' salaries and fringe benefits, fixed
by Congress; 15.5 percent for transportation
of mail, set by Government agencies; 1.8 percent for rents, utilities and communication
set by F.C.C. Only 4.3 percent is left for
controllable expenses over which the Department really excercises any control: printing
stamps, buying trucks, etc.
Three reasons why it is costing more to
give the American people the mail service they
deserve, expect, and pay for, are these:
1. More mail than estimated by about a
billion pieces. And this increased volume of mail doesn't provide funds to operate the Post Office Department, since
all revenues go directly to the Treasury
De par tmen t •

*
1'nq~,It1e$ ate invited
i, from tItO$. qualified as:

t • ELECTRICAL ENGINEERS '
:' ,'. ELECTROMECHANICAL ENGINEERS
" .' MECHANICAL DESIGN ENGINEERS
" • MECHANICAL ENGINEERS
, • MATHEMATICIANS
• PHYSICISTS

2. Increased mail service to 1.3 million
new homes and 250,000 new businesses.
3. Higher wages to employees as required
by law.
All three of these reasons are beyond control of the Department.
Are you cutting costs in the Post Office
Department?

A: Yes. The Department is now carrying
20 percent more mail with fewer employees than
six years ago.
Q: How many postal employees are there?
1952 ------------- 523,757
1955 ------------- 511,613
1956 ------------- 508,587

Q: How is mail volume increasing?
1952
1956
1957
1958

------------------(Est) ---(Es t) ----

49.9
56.4
58.8
59.7

billion
billion
billion
billion

pieces
pieces
pieces
pieces

(cont'd on page 44)

- 43 -

COMPUfER MARKET
(cont'd from page 17)

(2)

(1)

of Organization

(3)

(4)

Size

Recom.,
Title

(5)
Products and Services
(a)
(e)
(b)
(c)
(d)
Dig. Anal. Other Compo Servo

Code

T:iI~e

4T3

research

yes, -

4T4

insurances & annui ties

yes, pIng
off

4T5

crystal diodes &
transistors

no, -

4Ul

petroleum products

yes,
coordn

4Vl

comptg services

yes,
supvsr

4V2 education

yes, prof
of math

4V3

data on water resources

yes, proj
engr

4V4

advanced educ for naval
officers

yes, prof

4V5

aero controls

yes, res
supvsr

4V6

oil

yes, data
procg anal

4V7

communications

yes, supvsr
sys & proc

DMI'O

(6)
Likely Spending
Next Five Years
Between:
And:

$2,000,000

$3,000,000

2,000,000

4,000,000

MI'

PSO

200,000

250,000

500,000

1,000,000

500,000

700,000

DTO

2,000,000

3,000,000

250,000

500,000

1,300,000

1,500,000

v
v

0
M

*
U. S. MAILS

(cont'd from page 43)

.2,: Is the Post Office Department satisfied
with mail service today?

Forum

A: Of course not. It is better than it
was but still not good enough.

ASSOCIATION FOR COMPUTINi MACHIN!:RY MEETlNi
J. F. Summers
The Texas Company
P. O. Box 2332
Houston 1, Texas

.2,: What is the Post OffIce Department doing
to improve service?

A: It is spending over $4 million this
year alone on research. It has set up an
Office of Research and Engineering. It is
working with 9 engineering firms and the
Bureau of Standards. It is developing mechanical and electronic devices to speed mail
through post offices. Over 1500 obsolete
post offices have been replaced in the past
four years with new ones built by private capital. Our objective is next day delivery of
mail anywhere in the United States

The Twelfth Annual Meeting of the Association for Computing Machinery will be held at
Houston, Texas, on June 19-21, 1957, on the
campus of the University of Houston at Houston,
Texas. The program will be printed around
May 1, and may be obtained by writing Association for Computing Machinery, 1957 ,Meeting
Headquarters, University of Houston, Cullen
Boulevard, Houston 4, Texas.

- END-

*
-44-

* ------------------- * ------------------- *
RCA oilers opportunities

IN MISSILE TEST

Data Reduction
for
~ MATHEMATICIANS
~ STATISTICIANS
~ PHYSICISTS

~ ASTRO-PHYSICISTS

e·

Degree plus experience in reduction of
test data, applied mathematics, sta·
tistical techniques, or observatory
practices. Positions now available on
Florida's central east coast.
Liberal company benefits-Relocation
assistance.
For information and arrangements for
personal interview, send complete
resume to:
Mr. H. N. Ashby
Employment Manager, Dept. N-14E
Missile Test Proiect
RCA Service Co., Inc.
P.O. Box 1226
Melbourne, Florida

RADIO CORPORATION OF AMERICA

- 45 -

ADVERTISING
The purpose of COMPUTERS AND A UTOMA TION is to
be factual, useful, and understandable. For this purpose, the kind of advertising we desire to publish is
the kind that answers questions such as: What are
your products? What are your services? And for each
product: What is it called? What does it do? How well
does it work? What are its main specifications?

INDEX

R. C. A. Service Co., Missile Test Project, Patrick
Air Force Base, Florida / Employment Opportunities / Page 45, CA No. 16
Ramo-Wooldridge· Corp., 5730 Arbor Vitae St., Los
Angeles 45, Ca.JJL / Employment Opportunities /
Pages 2, 45 / CA No. 17
Sylvania Electric Products, Inc., 1740 Broadway, New
York 19, N. Y. / - / Pages 4, 5 / CA No. 18

Following is the index and a summary of advertisements.
Each item contains: Name and address of the advertiser
/ subject of the advertisement / page number where it
appears / CA number in case of inquiry (see note below).

READER'S
AMP, Inc., 2100 Paxton St., Harrisburg, Pa. / - /
Page 48 / CA No. 12
Burroughs Research Center, Paoli, Pa. / Employment
Opportunities / Page 43 / CA No. 13
General Transistor Corp., 1030-11 90th Ave., Richmond Hill, N. Y. / Transistors / Page 7 / CA No. 14
Lockheed Missile Systems Division, Box 504, Sunnyvale,
Calif. / Employment Opportunities / Page 47 / CA
No. 15

INQUIRY

produc~s or
services mentioned in one or more of these advertisements, you may circle the appropriate CA Nos.
on the Reader's Inquiry Form below and send that
form to us (we pay postage; see the instructions). We
shall then forward your inquiries, and you will hear
from the advertisers direct. If you do not wish to
tear the magazine, just drop us a line on a postcard.

If you wish more information about any

*----------------------------------------------*------------------------------------------------.
REPLY
-

READER'S INQUIRY FORM

envelope:~

Paste label on
-

Enclose form in envelope:·-.It
-

z to
...0 C
0

~G)
m

..,m
:::II::

m
rm

m
Z
-I
m

..,

Vi
m

CIt

~
z
m

-Z

(,I)
(,I)

I
I

- -

INQUIRY

.... -

FORM

Name (please print) ...........................................................................................................................
Your Address ? ......................................................................................................................................
Your Organization ? .........................................................................................................................
Its Address ? ........ '" ..................................... _..........................................................................
Your Title?.......... ................. ...... ......

. ................ " ...................................... .

Please send me additional information on the following subjects for which I have
circled th'e' CA number:

E=(JJ
.til m

1
6
11
16
21

?>r-

ZVl "'0

I
I

~!'

mCD

- - ... -

READER'S

2
7
12
17
22

3
8
13
18
23

4
9
14
19
24

5
10
15
20
25

26
31
J6
41
46

'0
32
37
42
47

28
33
38
43
48

29
34
39
44
49

30
35
40
45
~

51
56
61
66
71

52
57
62
67
72

53
58
63
68
73

54
59
64
69
74

55 76 77 78 79 80 101 102 103 104 105
60 81 82 83 s.4 85 106 107 108 109 110
65 86 87 88 89 90 III 112 113114115
70 91 92 93 94 95 116 117 118119120
75 96 97 98 99 100 121 122 123 124 125

126
131
136
141
146

1'0 128
132 133
137 138
142143
147 148

129 130
134 135
139 140
144 145
149. ISO

rrJ

::u

:!!

w
::a
VI
-<~ 3: -t
0::a-v
0
1'. Z
r5:

REMARKS:

111111111111111
1_

I
I
I
I

u: :::c
~ m
-<

'":;;

I
I

(,I)

~ m
CI:I

-1 ,,- -

________

0
z·
• lie
:< fO

VI
VI

O'l
CD

_ _ _ _ _ ..... _

..1_, ________________________________________ _I

- 46 -

MISSILE SYSTEMS ELECTl~ONICS
Activities at Lockheed Missile Systems Division laboratories
in Palo Alto cover virtually every field of electronics
related to missile systems. Inquiries are invited from those
who desire to perform research and development of a
most advanced nature.
Here staff members discuss a laboratory model of an airborne
component oj a guidance system. Left to right: Dr.R.J.Burke,
telemetering; E.A.Blasi, antennas; K. T. Larkin, radar and
command guidance; Dr.S.B.Batdorf, electronic division head;
Dr.H.H.Leifer, solid state; S.Janken, product engineering.

~IISSILE

SYSTE:\,IS DIVISION

research and engineering staff
LOCKHEED AIRCRAFT CORPORATION
PALO ALTO. SUNNYVALE • VAN NUYS
CAI ..IFORNIA

Design-Engineered with
Positive Wiping Contact
and
Frictional Grippage

~ C-~
TO

BETTER

Another of the many unique designs
made available by the AMP-Edge
technique is the new, low-cost, compact AMP-Edge Connector Block. It
allows freedom of arrangement,
with small area displacement.

Its open construction
provides aeration to
prevent moisture entrapment.

WIRINC

Source Exif Data:

File Type                       : PDF
File Type Extension             : pdf
MIME Type                       : application/pdf
PDF Version                     : 1.3
Linearized                      : No
XMP Toolkit                     : Adobe XMP Core 4.2.1-c041 52.342996, 2008/05/07-21:37:19
Create Date                     : 2015:08:24 15:21:57-08:00
Modify Date                     : 2015:08:24 14:40:21-07:00
Metadata Date                   : 2015:08:24 14:40:21-07:00
Producer                        : Adobe Acrobat 9.0 Paper Capture Plug-in
Format                          : application/pdf
Document ID                     : uuid:165eaff0-1047-464d-8335-ec4332f70c0c
Instance ID                     : uuid:f2f08d48-3a0c-0f45-a0b1-ff4b2b2363b8
Page Layout                     : SinglePage
Page Mode                       : UseNone
Page Count                      : 48

EXIF Metadata provided by EXIF.tools

195705

195705 195705

Navigation menu

Versions of this User Manual:

Views

Navigation