– 1 –

SETUN’s reflections

How the SETUN computer was perceived in the “Western”

scientific community

“The work day began with ‘morning exercises’: Each

employee of the laboratory, including the project manager,

got five ferrite cores of a diameter of 3 millimeter

…” (N. P. Brusenzov)1

1. Introduction

In 1958 the Russian engineer Nikolai P. Brusenzov and his small

team constructed the world’s first and still unique ternary computer.

The computer that was built at the Research Computing

Laboratories of the Moscow State University (MSU) was named

after a nearby small river, the SETUN’.

Although the iron curtain divided Eastern from Western scientist,

there was a huge interest on both sides to learn about each

other’s progress in technology.

The following research examines, how scientists of the “Western Block” learned of

the existence of the SETUN computer and how the information was reflected in

Western scientific publications.

It seems, that the existence of the SETUN in the Soviet Union at least supported and

sometimes even triggered the research in the “Western World” in the field of ternary

logic, a subset of multi-valued logic. Though its’ constructor can not claim to be

the first who thought about ternary logic and computing technology,2 the impact of

SETUN’s’ existence should not be underestimated. In a summary on the development

of multi-valued logic and computing technology given in 1977, Epstein/

Frieder/ Rine put it this way: “However, the SETUN computer awakened interest in

subsystems such as arithmetic units [Haberlin/ Müller 1970, Yoeli/ Halpern 1968,

Vranesic/ Hamacher 1971, Mine et. al. 1971] and numerous electronic modules as

cited in an abridgement of the bibliography for Thelliez’ doctoral thesis [Thelliez

1973].”

1 In: Malinovski 1995, [transl. F.H.]

2 Grosch, H. J. R.: Signed ternary arithmetic. Digital Computer Lab. Memo. M-1496, MIT, Cambridge,

Mass., May 1952

Img. 1: N.P.

Brusenzov (2004)

– 2 –

The 1958 SETUN’s’ appearance and the fact that scientist on the Western side of the

iron curtain learned about it in 19593 could be seen in the context of the Soviet’s

launch of the Sputnik Satellite in Oct. 4, 1957.

The Sputnik had a huge impact on the advancement of scientific research in the

United States and other Western countries. Or how the Historian Walter A. McDougall

put it: “No event since Pearl Harbor set off such repercussions in public life.”4

The American public was quite surprised that the Russians where the first to launch

a rocket and a satellite to successfully reach the earths’ orbit. Though the Americans

were aware of the importance of a space program and expected to launch their own

satellite soon (Project Vanguard - launch planned for November 1957) the American

public and with them scientists, military and the government were overwhelmed.

“There was a sudden crisis of confidence in American technology, values, politics

and in the military. Science, technology, and engineering were totally reworked and

massively funded in the shadow of Sputnik.”5

With the intention to catch up the technological and military gap,6 the Men on the

Moon Program was announced by John F. Kennedy in May 25, 1961 before the US

Congress with the words: “I believe that this nation should commit itself to achieving

the goal, before this decade is out, of landing a man on the moon and returning

him safely to the earth. No single space project in this period will be more impressive

to mankind, or more important for the long-range exploration of space; and

none will be so difficult or expensive to accomplish.“

In the consequence of this announcement, the technological research in the US was

generally boosted.7 The outcomes of the following basic research eventually lead not

just to space travel but also to technologies as the multimedia computer, Internet or

the Global Positioning System.

It is obvious that the SETUN didn’t make such a big public impression and didn’t

have such an impact on science. But in the light of what was said about the Sputniks

impact to science, it can be assumed, that SETUN’s pure existence has triggered or

at least supported the research that went on the Western side of the iron curtain.

3 Robertson, James in: Carr III, John W. 1959

4 McDougall, Walter A.: The heavens and the earth – A political history of the space age. (Basic Books,

New York, 1985) John Hopkins Paperback Edition, p142

5 Dickson, 2003, p 4

6 Walter McDougal argues, that a gap never really existed and that it rather was a successful production

of Soviet propaganda in connection with a lucky timing of the Soviet space program. (McDougal

1997)

7 Paul Dickson describes how the Men on the Moon Program changed the attitute towards science in

the US in the 1960’s. (Dickson, 2003 p 225 – 231 “Upgrading the three Rs”)

– 3 –

2. Information gathering

To understand, how the information on the SETUN was spread in the “West” (USA,

Canada, Great Britain, Israel and Japan), 44 out of 100 scientific documents were

examined for direct references to SETUN and/or its constructors. The research spans

a period of 20 years from 1958 to 1977 and included documents, which were published

in the US from US-American, Canadian, British, Israeli, Japanese, Argentine

and Indian authors. The 44 examined documents present the main body of still accessible

texts. 66 other documents could not be found or obtained from libraries for

various reasons. 8 The examined articles were found via the Columbia University Library9

and Compendex Catalogue10 either by cross-references or one of the following

keywords:

- three-valued logic

- ternary / trinary

- base-3

- TERNAC

- SETUN

- Brusenzov

While the indirect references where not counted,11 13 out of 44 documents where

found, that named the SETUN and referred directly to descriptions of the SETUN.

These references refer basically to 3 different instances:

1. Two separate groups of American scientist visited the Soviet computing centers

in 1958/ 1959 and actually saw the SETUN. Their published reports

were widely spread in the US in the following years. (Robertson et. al. 1959,

Ware et. al. 1960)

2. A summary on Soviet computing technology by Rudins 1970, drove its information

from Robertson 1959 and Ware 1960.

3. Some articles originally published in Soviet or Eastern Block publications in

Russian by Brusenzov or his co-developers were eventually obtained and

translated into English. Some scientists did translations personally; other

translations were reprinted in English in periodicals.

To give the reader an impression on how the SETUN was mirrored in the US academic

publishing in 1958-1978, all direct occurrences including the surrounding

context are cited below. Additionally comments are given, where found to be necessary.

8 For some documents, the references taken from the original articles, were not traceable in catalogues

and libraries at all. It can be expected that they exist somewhere but the time limit did not

allow to research all potentially accessible reference databases. For other documents the references

could be cleared, but lack of time, money or administrative problems circumvented the author to actually

get hold of them.

9 http://www.columbia.edu/cu/lweb/index.html

10 Compendex is one of the most comprehensive bibliographic databases of engineering research

available today, containing over eight million references and abstracts taken from over 5,000 engineering

journals, conferences and technical reports. http://www.engineeringvillage2.org

11 That means someone citing a document that contains a reference to a third document which names

the SETUN. From a raw review of the references, it can be assumed, that most of the documents have

at least one indirect reference to a text, where SETUN is named.

– 4 –

3. Sources and commentary

The first ever account of the existence of a ternary computer in the USSR was published

in June 1959 by Carr III, John W./ Scott, R. Norman/ Perlis, Alan J./ Robertson,

James E. in “A Visit to Computation Centers in the Soviet Union.” in the periodical

Communications of the ACM. Further references show that the gathered material

was also used for seminars and public lectures, spreading the information even

more.12

Author Affiliation

Carr III, John W. Research Computation Center, University of Michigan

Scott, R. Norman University of Michigan

Perlis, Alan J. Carnegie Institute of Technology

Robertson, James E. University of Illinois13

The visitors saw the SETUN and other Russian computers firsthand

but they did not mention Brusenzov at all, so it remains unclear if

they actually have met him personally. The article is a technical

description and gives just a few ideas, why the SETUN was constructed.

The “dissatisfaction with the STRELA”, a vacuum tube

based machine that was constructed during the same period at the

Moscow State University, was indicated by the author(s) as main

motivation for SETUN’s development. Unfortunately no reason for

Brusenzovs’ usage of three-valued logic was given.

Three out of the thirteen documents that name the SETUN are referring

to this article.

“The four authors spent a two-weeks period from August 27 through September

10, 1958, visiting Computation Centers in the Soviet Union at Moscow, Kiev, and Leningrad.

[…] They talked with Russian and Ukrainian computer mathematicians and engineers

working on comparable problems and were given very complete guided tours […].

The digital computer SETUN is under construction at Moscow University. More attention was

given to miniaturization than elsewhere among places visited, with elements mounted on cards

which were in plug-in trays perhaps 2" x 3" x 6" deep. The arithmetic unit, control, console,

and input-output control are mounted in one unit approximately 7' high by 11' long, with a

separate unit 6' long for the memory cores, the drum, and magnetic tapes. The lower 30" of

the units is not used.

SETUN is to be a base 3 machine with 18 digits per word, each digit having one of the values -

1, 0, or +1. The machine is to be serial, fixed point and with a single address in each of two

12 The article is often referred as “Carr 1959” pointing to John Carr who was the editor. James Robertson

actually gave the description of the SETUN, so the reference is used here as “Robertson 1959”.

Robertson 1959 In: Carr III, John W./ Scott, R. Norman/ Perlis, Alan J./ Robertson, James E.: A Visit

to Computation Centers in the Soviet Union. Communications of the ACM, v 2, n 6, 1959, p 8-10, p

14; Proceedings of the seminar on the status of digital computer and data processing developments in

the Soviet Union. ONR Symposium report ACR-37, Washington USA

13 Prof. James E. Robertson (1942-1999), an electrical engineer and expert in error-checking systems,

pioneers basic techniques of efficient binary division (now known as SRT). At that time the computers

IBM 650 and ILLIAC are installed at the University of Illinois.

Img. 2: John Carr

(undated)

– 5 –

commands per word. It was described as asynchronous, with a 200 kc. clock. Addition and

subtraction will require 180 µsec, multiplication 360 µsec. No division instruction will be provided.

A normalization instruction is included in the order code of 27 instructions to facilitate

floating point computation. One index register is provided with one digit of a command indicating

one of the three alternatives: add index, subtract index, or do not modify.

The memory hierarchy includes an 81-word ferrite core memory and approximately 2000

words of drum storage. The drum rotates at 7000 rpm, with a maximum access time of 14

msec, corresponding to two revolutions, one for waiting, one for transfer of a block of 27

words. The drum is physically small, perhaps 4" in diameter by 6" high, and has 60 tracks.

Addition of magnetic-tape units is planned at some later date.

Input and output will be on 5-hole punched paper tape; all transfers are to be in blocks of 27

words. The reader will be photoelectric, reading 400 lines/sec and requiring 15 to 20 lines to

stop: A printer is also to be available. Components are to be a “type of magnetic amplifier using

ferrite cores,” except for 70 vacuum tubes used as “generators” (drivers). The explanation

given for base 3 operations was that the hysteresis loop of the cores was not sufficiently square

and that compensation was required. Thus, two cores are necessary for each digit and three

states can be utilized. The motivation given for construction of the machine was dissatisfaction

with STRELA, said to be too complicated for open-shop university use.

A base 3 digit is stored in two cores in the core memory, on two tracks on the drum, and on

two holes in the paper tape. A single unit, containing one record tube and a five- transistor

read amplifier, can be switched to any one of three tracks of the drum.

[…]

It is the computer mathematics group at Moscow that is constructing the SETUN to provide for

simpler and more flexible operation of their center than is possible with the large and relatively

unreliable (10-minute mean free error time) STRELA. We were informed that the mathematics

group intended to continue research in computer design in the future. Thus, SETUN is a singleaddress

computer and algorithms are being developed which maximize the use of an accumulator

during extended calculation of algebraic formulae.” (Robertson in Carr 1959)

Just 8 months later in 1959, and after a Soviet scientist group

had visited the US; another group of US scientist was invited

to see computing machines at several sites in the USSR. Once

again the SETUN, which had been finished in December 1958

by Brusenzov and his team, was shown. This time his name

got mentioned. It is more than astonishing, that later on nobody

who cited this document mentioned Brusenzov as the

constructor of the SETUN and as the one, who had the idea to

use ternary logic.

The article by Ware, Willis. H. (ed.)./ Alexander, S.N./

Armer, P./ Astrahan, M.M./ Bers, L./ Goode, H.H./ Huskey,

H.D./ Rubinoff, M. got published as “Soviet Computer Technology”

in the periodical Communications of the ACM in

March 1959, then in the IRE Transactions on Electronic

Computer in Mar 1960, as a report for the RAND Corp as file RM-2541 in March

1960, and in the Bulletin Provisional International Computation Centre in July-

October 1960.

Img. 3: Willis Ware

(ca. 2000)

– 6 –

Author Affiliation

Ware, Willis. H. RAND Corp.

Alexander, Samuel N. National Bureau of Standards, United States

Armer, Paul RAND Corp.

Astrahan, Morton M. IBM

Bers, Lipman Institute of Mathematics, New York University

Goode, H.H University of Michigan, Bendix System Division

Huskey, H.D. University of California

Rubinoff, M. Philco Corp.

“ Thursday, May 21 [1959]

The rest of the group [Astrahan, M. M./ Alexander, S. M./ Armer, P./ Bers, L./ Huskey, H.D./

Ware, W. H.] visited the Lomonosov Campus of the Moscow State University and its computing

center. The work of this computing center was discussed and the STRELA and SETUN installations

were visited and described. We met or spoke with the following:

Academican S.L. Sobolev, Head, Computing Chair, MSU […]

N. P. Brusenzov, Chief engineer of the SETUN machine

[…]

SETUN

This base-3 machine being constructed at Moscow State University appeared to be in operation

when we saw it (fig. 16). It was explained that the choice of base-3 was made because it can

be shown that in some sense a base of 3 provides the most efficient utilization of equipment.

[Footnote missing] Since a base-3 electronic technique is not available, they decided to construct

a base-4 machine and to utilize only 3 of the 4 possible states. The unused 4th state in

each case is available for some form of checking. This machine is regarded as experimental,

and as an educational training program for engineers. In part they felt that SETUN was a protest

against the huge, complicated machines being built elsewhere. It was thought easier to operate

a simple machine at their center. The machine contains 4,000 magnetic cores, 4,000

germanium diodes, approximately 100 transistors, and 40 vacuum tubes. It operates at a 200-

kilocycle clock rate. It uses 1 MC transistors, which are rated at 150 milliwatts dissipation at

25 degree centigrade, but can tolerate a maximum of 100 degree centigrade.

SETUN has only 81 words of storage and 27 different instructions. It is a single-address, fixedpoint

machine, with 18 ternary digits per word. The point is fixed between the second and

third digits from the left. It is serial and contains two instructions per word. There is no divide

instruction.

The ferrite core store can be regarded as having 162 9-digit words because the half-words can

be addressed. The drum store contains 2,268 half words. Number representation of SETUN requires

2 binary digits per base-3 digit; therefore a 9-digit, base-3 word, will require 18 tracks

on the drum. There are three such groups of 18 tracks on the drum, or a total of 54 heads. In

each band of 18 tracks there are 756 words recorded in parallel; there are thus 756 bits

around the 13-inch circumference. It is planned to add magnetic tapes to this machine at some

later date.

– 7 –

Addition time is 180 microseconds, including all accesses. Fetching of the next instruction is

overlapped with the execution of the previous one. Multiplication time is 335 microseconds,

and transfer of control is 100 microseconds. SETUN includes a normalizing instruction (shifting

operations to facilitate programming of floating-point), one index register, and teletype input

and output. The German type RFT teleprinter is partially base-9 and partially base-3. A 9-digit

word is printed as two base-9 digits, then one base-3 digit, then two more base-9 digits. The

characters used are:

Base-9:

Base-3: i,0,1

Five-level punched paper tape is used for the input and output.” (Ware et. al. 1959)

This seems to be even more detailed information than in Robertson 1959 and it can

be assumed the Robertson’s’ account has been used to verify and complete the document

by Ware et. al. 1959.14 Now we get some additional information on SETUN’s

context: “This machine is regarded as experimental, and as an educational training

program for engineers. In part they felt that SETUN was a protest against the huge,

complicated machines being built elsewhere.” Instead of pointing out the STRELA as

in the former article, it was more generally referred to “complicated machines elsewhere”

as an argument of distinction for the SETUN. No author was given for this

statement. Additionally the question, why ternary logic was used, remained unanswered

again.15

The next article – published 2 years later, in 1961 – was based on the research of

Hallworth, R.P./ Heath, F.G. from the University of Manchester. They researched

about “Semiconductor circuits for ternary logic” and don’t name the SETUN directly

but give a clear reference to Robertson 1959.

Author Affiliation

Hallworth, R.P. IBM British Laboratories, former Member of Faculty of Technology,

University of Manchester

Heath, F.G. Faculty of Technology, University of Manchester

14 It is worth to mention as well, that Willis H. Ware, the editor of this article, at that time was an

employee of the RAND Corporation.

Industrialist Donald Douglas approached the Army Air Force in January 1946 with a plan for joint

industry-government coordination – in short, a think tank. Project RAND (coined by Arthur Raymond

from Research and development) was originally founded with Douglas Aviation. In May 14, 1948

Project RAND separated from Douglas and became an independent, nonprofit organization. The think

tank established a special field of sciences, called “Soviet Studies” about ten years later, following the

Sputnik shock. „In a real sense, Soviet studies were invented at RAND.“ (Jonathan D. Pollack). This

departement collected and examined material from various sources (unclassified documents, news

sources, intelligence services etc.) about the social, sociological, political and scientific life of the Soviet

Union which was delivered in public or classified publications to a variety of readers in the US

and the Western world. The goal was to better understand recent Soviet developments and to enable

decion makers to react in a short time.

15 Brusenzov later argued, that a general advantage of ternary logic was the “most natural number

system” (Brusenzov) and also mentioned the demand to reduce the amount of components to achieve

a relatively small construction. [Malinovski 1995, Rumjanzev 2004]

– 8 –

“Interest has increased recently in non-binary switching theory, because certain advantages

may be obtained by using a radix higher than 2 in digital computers [Robertson 1959] as well

as digital communication systems. […] Generally in both applications binary circuits and logic

have been used because of their simplicity. It can be shown, with certain assumptions, that the

most efficient radix for computing is 8, which makes 3 the best integral value.” (Hallworth/

Heath, 1961)

R. D. Merrill was one of the most active authors in this field, 7 publications from

1963-1973 on ternary logic were found.16 Merrill was the first to give another reference

to SETUN than Carr 1959 and Robertson 1959. He cited an article on “The order

code and an interpretative system for the SETUN computer.” from 1962 by E.A.

Zhogolev, who was the main programmer in the SETUN team.17

Author Affiliation

Merrill, R. D. Electronic Sciences Lab, Lockheed Missiles & Space Co., Palo

Alto, Calif., USA

“The SETUN, a Russian stored program computer, incorporates the ternary number system in

its arithmetic unit. [Zhogolev 1962] The mechanization was not wholly ternary since the control

logic is binary and information is binary encoded for storage in the bulk and random access

memories. Evidently the SETUN was so designed to take advantage of the relative ease of representing

negative and positive numbers and the simplicity of performing round-off when using

the ternary representation.” (Merrill 1965)

When Merrill stated: “Evidently the SETUN was so designed to take advantage of the

relative ease of representing negative and positive numbers and the simplicity of performing

round-off when using the ternary representation.”, he was giving for the

first time a possible reason for the use of ternary logic. His use of “evidently” shows,

that he expressed his point of view, not Zhogolevs.18 One year later, in 1966, Merrill

puts the SETUN in a broader context.

“Moreover there have been several instances reported recently where ternary logic and switching

networks have found practical application. For example SETUN, a Russian stored-program

16 See Appendix A

17 “I will check my files, but seem to remember that this information came from [Antonin Svoboda]

from Czechoslovakia, in this country then, who was famous at that time for his work with the application

of the Chinese Remainder Theorem (modular arithmetic) to digital computing and error correcting

codes. His work was given to me by Dr. Howard Aiken, then the director of Harvard University's

Laboratory for Computing.” (Merill, Email 30.3.2005), Svoboda built the first Czechoslovakian computer

named SAPO in 1957.

18 Zhogolevs introduction reads: “However developed the order code of a computer may be, it provides

the mathematician with a very imperfect tool to describe the computational processes. This tool

can be considerably enlarged and improved with the ususal methods for making programming more

automatic (standard sub-routines, compiling and interpretive systems, programm generators, etc.) The

need for an improvement in the basic programming structure is most acute for small computers with

a simple logical structure and, usually, a small set of very elementary operations.

In this article we consider the first steps towards an improvements of the basic programming of the

SETUN unsing an interpretative system which can be a bsasis for the future improvement of this apparatus.”

[Transl. by R. Feinstein] (Zhogolev 1962)

– 9 –

computer, incorporates the ternary number system in its arithmetic unit to take advantage of

the relative ease of representing negative and positive numbers and the simplicity of performing

round-off in the ternary representation [Zhogolev 1962, Merrill 1965]. Also many proposed

content addressable memories utilize tristable switching and memory elements to accomplish

masking and associated operations [Lewin 1962]. Ternary switching theory has been proposed

as a useful means of designing hazard-free binary gating, and series parallel contact networks

[Eichelberger 1965, Yoeli/ Rinen 1964]. NASA has actively supported the development of tristable

fluid logic devices for hydraulic switching systems [Reader/ Quigley 1963/63]. Still another

area is the use of ternary logic redundancy in binary networks to improve operating reliability

[Varshavsky 1964].” (Merrill 1966)

This overview given by Merrill shows how the field of research on ternary logic and

computation has broadened after 1958. And this was what Yoeli/ Halpern stated as

well in their article from 1968: “Only recently, considerable interest in non-binary,

and especially ternary, switching circuits has arisen owing to the potential advantages

of ternary over binary systems.“

Author Affiliation

Yoeli M. Faculty of Electrical Engineering, Technion, Haifa, Israel

Halpern, I. Faculty of Electrical Engineering, Technion, Haifa, Israel

“Only recently, considerable interest in non binary, and especially ternary, switching circuits

has arisen owing to the potential advantages of ternary over binary systems. […]

The Russian computer SETUN [Brusenzov 1962, Brusenzov 1965] is an interesting experiment

of a digital computer, which uses a symmetric ternary number representation. This number

representation has many advantages, which are discussed in the sequel. However, whereas the

SETUN uses magnetic circuitry in its arithmetic unit, this paper proposes a ternary arithmetic

unit which supplies present-day diode-transistor circuitry. Detailed circuit realization, and logic

diagrams of ternary gates, of a 3-stable element, and of a full adder are developed. From these

modules, a complete ternary arithmetic unit can be simply constructed.” (Yoeli/ Halpern

1968)

Yoeli and Halpern are the first to directly cite from articles written by Brusenzov.19

They see the SETUN as “an interesting experiment” which in the broader context of

general computer development was an obvious consideration. Yet, Brusenzov himself

stressed the fact that the SETUN was used in 30 installations spread all over the

USSR and that the users were satisfied with its function and reliability. So in his

opinion it was more than just “an interesting experiment”. [Malinovski 1995, Rumjanzev

2004]

D.I. Porat referred in his 1969 article “Three-valued digital systems” to Robertson

1959 without addressing the SETUN directly and rather talking of the existence of a

“ternary computer” in general. As in many other articles the reference to the SETUN

was given to support the own research.

19 In an email to the author Michael Yoeli writes: “I don't remember how I came across the references

you mention …” (Yoeli, M. 12-Mar-05) and regarding his knowledge of Russian language: “I

did take a beginners' course in Scientific Russian, but did not make much progress.” (Yoeli, M. 18-

Mar-05)

– 10 –

Author Affiliation

Porat, D. I. Stanford University, Calif. USA, former at University of Manchester,

UK

“Digital equipment design is based on the binary-number system because of the availability,

low cost and reliability of binary switching and storage elements. Higher radix systems are

implemented by use of binary coding; however, at least one ternary computer [Robertson

1959] has been in operation which incorporates 3-valued elements.” (Porat 1969)

Author Affiliation

Rudins, George Rand Corp

A short summary on the SETUN was given by George Rudins in his article “Soviet

computers: a historical survey.” It was published in RAND’s Soviet Cybernetics Review

in January 1970. The aim of the journal was “to disseminate to a wide range of

specialists information about Soviet publications, activities, and new developments

in computing technology, cybernetics and scientific policy.” (p.ii). The research was

supported by the United States Air Force under the Project RAND-Contract No.

F44620-67-C-0045. What is astonishing is the fact the Rudins, who was the managing

editor of Soviet Cybernetics Review, didn’t give any references for the complete

article. It is most likely that he drew his information from Robertson 1959 or Ware

1959.

“The first Soviet computer with alphanumeric I/O capabilities, the Setun’, was developed in

1958 as part of a graduate student project at Moscow State university; N. Brusenzov, who

worked on the Strela with Basilevskij, was also involved in this project. The Setun’ was capable

of performing 4000 opns/sec (1-adress), and had 81-18-bit words of core store. It was the

world’s only computer to ever use base-3 logic. According to the Soviets, base-3 logic was supposed

to provide the most efficient utilization of hardware. Since base-3 electronic technique is

nonexistent, they decided to construct a base-4 machine and to utilize only three of the four

possible states. Although the entire project was regarded as an educational training program for

engineers, an attempt was made to serially produce it, but it failed miserably – base-3 logic

turned out to be highly impractical.” (Rudins 1970)

From nowadays-available information it becomes obvious, that Rudins’ summary contains

wrong and incomplete information: Brusenzov never worked with Basilevskij

on the STRELA computer, as the article supposes. They rather worked within the

same organizational structure at the Moscow State University. The STRELA was produced

at the newly established Computational Center at the MSU in the Special

Construction Office SKB-245. Brusenzov's laboratory was also part of the Computational

Center but not directly connected to the SKB-245.

Rudins also mentioned that an attempt was made to produce the SETUN serially.

Obviously his information was not precise as well, when he wrote: “an attempt was

made to serially produce it, but it failed miserably” It is true that mass production of

the SETUN failed. There was for instance the plan to produce some thousand copies

a year in the Peoples Republic of Czechoslovakia, which was turned down by the Soviet

authorities because they wanted the money to be earned by themselves rather

than by the Czechs (Brusenzov in Rjumanzev 2004). But it is not true that the serial

production, as Rudins claims, failed. There was a serial production of the SETUN at

– 11 –

the Kazan Mathematical Machines plant20 starting in November

1961 and releasing 50 copies which were spread all over the

Soviet Union (Malinovski 1995).21 In this context it becomes

obvious, that also Rudins conclusion “base-3 logic turned out to

be highly impractical” is shortening if not even wrong.22

Gideon Frieder started to work on the TERNAC in the early

1970s. His goal was to create a ternary computer for evaluation

of ternary logic and arithmetic against binary logic. The TERNAC

wasn’t a computer itself but rather an emulation written

in FORTRAN of a ternary computer on another machine, a Borroughs

B1700. He cites Ware 1959 for his information about

the SETUN.

Author Affiliation

Frieder, Gideon State University of New York/ Buffalo, USA

Luk, C State University of New York/ Buffalo, USA

“There seems to be quite an abundance of hardware units built for ternary computers, including

adders, multipliers etc. [Dept. of EE, 1971]. Furthermore, a ternary computer was built in

the University of Moscow in the late 50’s [Ware et. al. 1959]. These existing devices illustrate

exactly the problem to which we referred in the previous passage; they employ base-3 arithmetic

without any reference at to how such arithmetic should be applied. Although ternary systems

were known for a while, and in particular balanced ternary was looked into [Knuth,

1969, vol. 2, pp.173-176], there is, to the best of my knowledge, no a priori attempt in those

systems to decide if arithmetic should be implemented in balanced or regular ternary.”

(Frieder/ Luk, 1972)

In 1974 Gideon Frieder gave “A Summary of the Development of

Multiple-Valued Logic as related to Computer Science.” at the

Symposium on Multiple-Valued Logic in Cooperation with G. Epstein

and D.C. Rine.23 They refer to two sources regarding the

SETUN. Again 196024 is mentioned for the first time as a reference

and Zhogolev 1962 to who was first referred by Merrill 1965.25

20 Kazansk Mathematical Machines Factory (now: ICL), 34 Sibirsky Trakt, 420029 Kazan,

http://www.icl.kazan.ru/eng/activities/production/background/

21 For comparision: DEC produced 50 copies of the fully transistorized PDP-1 from 1961-1964 (Ceruzzi

1998, p 128)

22 For further information on the Soviet computer technology industry and the state system of planning

and strategy see Klimenko 1999.

23 “SETUN actually started my interest in MVL [Multiple Valued Logic – F.H.] from a digital systems

point of view, and we had read the early works of SETUN in the early 1970's.” (Rine 23-Mar-2005)

24 I’ve found one article by Again 1960 but it doesn’t address SETUN directly.

25 See also Appendix A.

Img. 4: Gideon

Frieder (ca. 2000)

Img. 5: D.C.

Rine (ca. 2000)

– 12 –

Author Affiliation

Frieder, Gideon IBM Israel

Epstein, G. Indiana University

Rine, D. C. Department of Statistics and Computer Science, West Virginia

University, Morgantown, USA

“The earliest document seriously proposing a full scale ternary computer was written in 1952

by Grosch [Grosch, 1952]. It advocated the implementation of a balanced ternary (digits –

1,0,1; radix 3) arithmetic unit for the Whirlwind II computer which was then in design.

The first full scale implementation of a ternary computer called SETUN was completed in 1958

at Moscow State University [Again 1960, Zhogolev 1962]. Restricted by poor hardware reliability

and inadequate software, there was no extensive attempt to use SETUN for critical comparison

of binary and ternary computers in the area of arithmetic.” (Frieder/ Epstein/ Rine 1974)

There is an interesting turn in the context, which they draw for the development of

using ternary logic and computing technology. When they mention Grosch and his

thoughts about ternary logic for the WHIRLWIND computer in 1952, it shifts the

mentioning of the SETUN to be the first ternary computer a little bit toward the fact,

that even before1958 someone was at least thinking about ternary computing. On

the background of the Sputnik shock in 1957 it is obvious, that at least some importance

was given to the question: “Who was the first?”

Frieder/ Epstein/ Rine stated further, that the SETUN was “restricted by poor hardware

reliability and inadequate software” which at least if one believes Brusenzov

was not the case. He especially he stressed the high reliability of the SETUN (90 percent

usage time) in comparison to other Soviet computers of that time. (Rjumanzev

2004). Advancing to it from the point of view some 15 years later and from the own

interest in comparing ternary and binary computing – something the SETUN was not

intended to be for – the authors somehow blur the reason and the cause when stating:

“Restricted by poor hardware reliability and inadequate software, there was no

extensive attempt to use SETUN for critical comparison of binary and ternary computers

in the area of arithmetic.”

The author of the following article, Z.G. Vranesic, was one of the

most published authors in the field of ternary logic.26 He researched

at that time at the University of Toronto. Although his

first traceable article on the subject is dated to 1971, it takes three

more years until he refers to the SETUN computer. When he finally

referred to SETUN in 1974 he didn’t cite from the former

known sources but rather reveals a new original Russian article

from the Moscow State Universities Vestnik (1962).27

26 See Appendix A

27 “I don't remember where I found the reference to Brusenzov's paper, but I got a copy of it through

the University of Toronto library. The paper was in Russian, which I can read because I grew up in

Croatia (which was then a part of Yugoslavia) and had to take Russian as a required foreign language.

The paper was interesting only as one of the first attempts to build a non-binary computer.” (Vranesic

14-Mar-2005)

Img. 6: Z.G.

Vranesic

(ca. 2000)

– 13 –

Author Affiliation

Vranesic, Z.G. Department of Electrical Engineering and Computer Science,

University of Toronto, Ontario, CA

Smith, K. C. Department of Electrical Engineering and Computer Science,

University of Toronto, Ontario, CA

“Some of the early attempts concentrated on the development of basic devises which were essentially

non-binary in nature. Such was the case of the Rutz transistor [Kaniel 1973], the parametron

[Schauer/ Steward/ Pohm/ Reid 1960], multi-aperture square loop ferrite devices,

etc. [Anderson/ Dietmeyer, 1963]. Magnetic cores received special attention [Ivaskiv

1971,Santos/ Arango/ Pascual/ 1965] and even became a key building block in the SETUN

computer [Brusenzov/ Zhogolev/ Verigin/ Maslov/ Tishulina 1962]. Eventually these attempts

gave way to the approach of utilizing readily available binary components for construction of

circuits which exhibit multi-valued behavior.” (Vranesic/ Smith 1974)

It is not clear, where exactly Rath has taken his knowledge of the SETUN. In the

general references to his article he named Hallworth/ Heath (1961), Merrill (1966),

Porat (1969) and Yoeli/ Halpern (1968) who all refer to the SETUN computer and

partly also give a small description of it.

Author Affiliation

Rath, Shri Sudarsan Department of Electrical Engineering, Regional Engineering College,

Rourkela, Orissa, India

“Ternary means an element of a switching system which will perform 3-valued transmission

and 3-valued switching. The Russian computer SETUN is an interesting experiment in digital

computation. It uses 3-valued devices and follows a symmetric ternary system of logic states

1,0,-1.” (Rath 1975)

In 1977 Epstein/ Frieder/ Rine contributed a chapter on The development of multiple

valued-logic as related to computer science in the Book Computer Science and

Multiple-Valued logic, theory and applications, edited by D.C. Rine.

Author Affiliation

Frieder, Gideon IBM Israel

Epstein, G. Indiana University

Rine, D. C. Department of Statistics and Computer Science, West Virginia

University, Morgantown, USA

“Ternary systems for computation were discussed in 1840 [Cauchy 1840, Lalanne 1840]. By

1950, there were ternary devices widely enough known to be included in a review book [Epstein/

Horn 1974]. However, it was not until 1958 that the first full scale ternary computer

was completed at Moscow State University [Again 1960, Brusenzov 1960, Robertson et. al.

1959, Zhogolev 1962] although ternary computers where proposed as early as 1952 [Grosch

1952]. While this encouraged work on subsystems such as arithmetic units in Canada [Vranesic/

Hamacher 1971], Japan [?], Switzerland [Haberlin/ Müller 1970], and Israel [Yoeli/

– 14 –

Halpern 1968], the low-level programming language devised for this Russian computer was so

difficult to use that there was little insight into 3-valued logic by users.

[…]

The earliest document seriously proposing a full scale ternary computer was written in 1952 by

Grosch. It advocated the implementation of a balanced ternary (digits –1, 0, +1) arithmetic until

for the Whirlwind II computer which was then in design. However, there was no discussion

in this document of algebraic or logic operations.

In the late 1950’s the first full scale implementation of a ternary computer as undertaken by a

Russian team at the Computing Centre of Moscow State University. This computer, completed

in 1948 and named SETUN, was briefly described by Carr et al. [Robertson/ Carr 1959] in a

1959 survey of Russian computers. It was used for some time but both poor hardware reliability

and inadequate software hampered its usage [Zhogolev 1962]. Additional details may be

found in [Again 1960, Brusenzov 1960].

The SETUN computer was a fixed-point arithmetic computer with words of 18 ternary units

(trits) in length. There were two trits to the left of the fixed-point rather than one. The memory

consisted of a core storage unit of 81 words and a magnetic drum unit of 1944 words.

The representation of a trit in flip-flops or the core storage unit was accomplished through the

use of two coupled cores, which provided three stable states. The hysteresis loops for these

coupled cores did not allow four stable states. The representation of trits on the magnetic drum

or input/ output tapes was through binary-coded ternary on two stacks, hence wasting a single

binary digit (bit) of information per trit. This waste was compounded through the use of a ternary-

decimal (a further waste of 17 its of information per decimal).

The SETUN was an arithmetic machine. The only logic operation it had was digit-by-digit conjunction

(logical multiply). It was therefore completely unsuitable for any evaluation of ternary

logical operations. Yet there was no extensive attempt to use it for critical comparison of binary

and ternary computers with respect to arithmetic operations.

However, the SETUN computer awakened interest in subsystems such as arithmetic units

[Haberlin/ Müller 1970, Yoeli/ Halpern 1968, Vranesic/ Hamacher 1971, Mine et. al. 1971]

and numerous electronic modules as cited in an abridgement of the bibliography for Thelliez’

doctoral thesis [Thelliez 1973].” (Epstein, G./ Frieder, G./ Rine, D.C. in Rine 1977)

This seems somewhat friendlier towards the SETUN than in their summary given in

1974, maybe also because they put it into an even broader context. Regarding the

usage and reliability they stick to their findings, which they made three years before.

Remarkable was their understanding, that the SETUN awakened the interest in the

further work on ternary computing devices.

It would be interesting to follow the development of the field in the beginning 1980s

with the broad access to microchips, which abandoned the use of transistors for

computer devices. There were unfortunately not enough resources to do further research.

So the chapter by Rine/ Frieder/ Epstein might be taken for the closing

chapter of the activities in the 1970s.

– 15 –

4. Why the research in the “West” eventually stopped.

The examination of these sources naturally evocates the question why the research

interest on ternary computing slowed down after a period of twenty years. The major

authors were contacted by e-mail and asked the questions: “Can you remember

where you got the knowledge of SETUN's existence from.” And: “Why did you stop

researching the field of ternary computing or ternary arithmetic?”

Michael Yoeli who was involved in ternary logic in the 1960s at the Technion in

Haifa comments the development in that time and SETUNS influence from his nowadays

point of view:

“We were attracted by the general idea of ternary computing. However, the problem is to design

the relevant hardware, in such a way that the ternary computer becomes competitive with

available binary computers. SETUN was an effort to deal with this problem, but there is no

doubt that this effort failed and the idea of the SETUN was indeed eventually abandoned. At

the time the SETUN was built in Moscow, computer hardware technology was just in its beginning.

However, even more advanced technologies do not offer a reasonable challenge. Our

computing unit was just a design on paper, but no effort was made to implement this idea.

From a software viewpoint the idea of ternary computing is indeed attractive, and the SETUN

was an important contribution to this idea.” (Yoeli 12-Mar-2005)

So Yoeli puts the success of binary memory units as a major reason, which made the

effort to actually transfer the theory of ternary logic into real applications, i.e. ternary

microchips, unattractive and unaffordable.

Roy D. Merrill who in the 1960 was associated with the Lockheed Missiles and

Space Corp Research Lab gives another viewpoint.

“We [Lockheed Missiles and Space Corp] had a contract with the Air Force to explore a number

of areas that might enable the computer designer to devise means of making the computer

operate faster and with less complexity. […] We researched Residue Arithmetic because addition,

subtraction and multiplication arithmetic operations could be carried out without waiting

for the carry bit. We researched ternary arithmetic because it could possibly have helped ease

the complexity of the logical design. Neither proved to be successful for other than very special

applications.” (Merrill, 26-Apr-2005)

Zvonko Vranesic, who was with the Department of Electrical Engineering and Computer

Science at the University of Toronto in the early 1970s, writes about the advantages

and disadvantages of ternary logic:

“Ternary logic has some useful arithmetic properties, such as allowing the balanced representation

of numbers. Its main drawback is that there do not exist any practical implementations of

ternary storage circuits. Thus, if one has to use binary storage devices, it is clear that it makes

more sense to try 4-valued logic circuits which can interface to the binary memory more efficiently.”

(Vranesic 14-Mar-2005)

And he further explains:

– 16 –

“The reason why storage devices are binary is that it is easy to use transistors as simple

switches that have to states, OPEN and CLOSED, which can be interpreted as 1 and 0. There

are no simple elements that can be implemented using the Integrated Circuit technology which

would naturally exhibit 3 states.” (Vranesic 17-Mar-2005)

This correlates with the account D.C. Rine gives on this question. He researched

during the 1970s at the Department of Statistics and Computer Science of the West

Virginia University in Morgantown, USA. Rine answers that the interest in multivalued

logic (of which 3-valued logic is a subset) had to do with the general development

of microprocessors. In the 1970’s companies such as IBM, Motorola, National

and others started to explore the combination of 2-state logic with 4-state logic

in microprocessors to improve speed and memory capacity. Applying the pure scientific

findings in multi-valued logic to the actual engineering process wasn’t always

easy.

“As we studied, and invented, the idea of Technology Readiness (TR) models and Technology

Maturity (TM), it was discovered that there is a transition for most technologies through 9 different

TM/TR stages from research ideas inception through integration into existing systems.

So it took the research inception of n-state logic28 to make its way from purely research (isolated)

ideas to a maturity and readiness such that the existing computer systems technology

could accept or interface with 4-state VLSI maturing technology. The very important point to

observe is that an important point in time much be reached when technologies evolving from

reach to more mature/ ready technologies can usefully interface with current existing computer

systems wide technologies. If evolving/maturing new technologies do not reach that

point in an appropriate time then they will generally never be used in current/existing systems.

Many interesting and novel research ideas never go beyond the phases from inception to

stand along maturity, and are therefore never used in existing systems.” (Rine 23-Mar-2005)

In the case of the 1970’s microchip development it turned out, that the merging of

2-state and 4-state logic was successful. But it turned researchers – and among them

Rine – away from further research in any n-state logic where n would be an uneven

power of 2.

“I turned my attention to directing PhD research dissertations using interval logics, formal

methods proof systems and formal natural logic communication systems which seemed to hold

more promise for my own research. And I could not see a way or a path for reaching the higher

TM/TR levels with 3-state VLSI29 logic. More recent dissertations/ papers appeared in those

areas.” (Rine 23-Mar-2005)

Gideon Frieder who led the development of the TERNAC emulator at the SUNY Buffalo

during the 1970s answers the question, why the research eventually stopped:

“Essentially, we were victims to our own success. In developing the algorithms for the emulation

of a balanced ternary machine on a binary host, we found that one can implement both

the [ternary] arithmetic and the logic on binary hardware. So if somebody is interested in using

a ternary computer, it is easy to implement it as software. The realities of Moore’s law showed

28 Multi-valued logic

29 Very Large-Scale Integration – Placing thousands of transistors in a small space – the microchip.

– 17 –

us that ternary hardware is not cost effective in the state of affairs in the 80s.” (Frieder 15-Jun-

2005)

To add a little bit of context to Rines and Frieders answers, the time period, the beginning

1960’s in which the SETUN was developed, should be considered as well.

When the first talks started about it at the Moscow State University recently founded

Computing Research Center started in 1956, the machine that would become the

SETUN was still thought of as being a binary computer. But at this time computing

technology in the Soviet Union was still in its early phase, which meant it was quite

experimental.30 Not much experience existed and many machines were built “from

the scratch”. For the SETUN for instance a seminar gathered for a few months, consisting

of mathematicians and electrical engineers and would make up a general

plan, of how to build such a calculating machine. Brusenzov was a radio engineer by

education. Working diverted from others and in parallel to other ongoing projects

(e.g. STRELA, BESM) on his own, it was relatively easy to come up with something

basically different as ternary logic.

When the research on ternary computing seriously started in the “West” in the

1960s, the situation and context was entirely different. A broad experience in building

binary computers and elements existed along a growing industry that was about

to switch from using ferrite cores and vacuum tubes towards transistors31. So it was

much more difficult to not just start a project on ternary computing but to actually

finish it in the sense, that a computer based on ternary logic would be build and

used.

The reasons for the research in ternary computing, ternary arithmetic and ternary

devices following the given accounts can be summed up as:

- general interest in ternary computing in the early, experimental years of computing

- use of balanced ternary numbering system (-1 | 0 | 1), especially for residue

arithmetic

- hope to simplify logical design

The set of reasons for the failure of developing a working ternary computer in the US

or Canada is closely connected to technology readiness and technology maturity:

- problem to design the relevant hardware

- problem to translate design on paper into working devices

- competing with existing binary hardware

- research in a complex field of scientific and/or military applications.

30 This experimental phase of computing was delayed a little bit compared with the US because universities

and plants in the Soviet Union had to be rebuilt after World War II and the human losses also

affected the sciences. Brusenzov, who was a radio operator during the war, was taken into the Soviet

Army in 1943 and finishing his 10th grade was delayed until 1948. Of the 25 young men, who entered

the war in 1943 only 5 , including Brusenzov, survived.

31 Cerruzzi 1998, p 64

– 18 –

It is obvious that technology doesn’t become into existence “out of nowhere”. There

is a historical, political and personal context that has to be considered. The article

tried to follow the “carrier” of the Russian SETUN computer through early Western

computer science. It could be shown, that the existence of the SETUN computer

partly triggered awareness in the “West” or at least supported the ongoing research

in form of an argument for own academic research projects and funding.

– 19 –

Bibliography

This bibliography gives only the references that were used in the above text and

cites. To see a list of all examined documents, please see Appendix B.

Again, J.: Setun. Nauchno-Tekhn Obchestva SSSR, v 2, n 3, 1960, p 25

Anderson, D.J./ Dietmeyer, D.L.: A magnetic ternary device. IEEE Transactions on Electronic Computers,

v EC-12, December, 1963, p 911-914

Brusenzov, N. P.: A ternary arithmetic machine. Moscow State University Vestnik, v 2, 1965, p 39-48

(in Russian)

Brusenzov, N. P.: The Computer Setun of Moscow State University, Material of the Scientific-

Technical Conference: New Developments in Computational Mathematics and Computing Techniques,

Kiev 1960, p 226-234

Brusenzov, N.P./ Zhogolev, Ye. A./ Verigin, V.V./ Tishulina, A.M.: The SETUN small automatic digital

computer. Moscow State University Vestnik, v 4, 1962, p 3-12 (in Russian)

Carr III, John W./ Scott, R. Norman/ Perlis, Alan J./ Robertson, James E.: A Visit to Computation

Centers in the Soviet Union. Communications of the ACM, v 2, n 6, 1959, p 8-10, p 14; Proceedings

of the seminar on the status of digital computer and data processing developments in the Soviet Union.

ONR Symposium report ACR-37, Washington USA

Cauchy, Comptes Rendus, n 11, Paris, 1840, p 789-798

Ceruzzi, Paul: A history of modern computing. MIT, 1998

Davis, N. C./ Goodman, S. E.: The Soviet Bloc’s Unified System of Computers. ACM Computing Surveys

(CSUR), v 10, n 2, 1978

Dept. of Electrical Engineering: Conference Record of the 1971 Symposium on the theory and applications

of multiple-valued logic design, May, 1972, State University of New York at Buffalo

Dickson, Paul: Sputnik – The shock of the century. Berkley Book 2003 (2001)

Eichelberger, E.B.: Hazard detection in combinational and sequential switching circuits. IBM Journal

on Research and Development, v 9, n 3, 1965, p 90-99

Epstein, G./ Frieder, G./ Rine, D.C.: The development of multiple valued-logic as related to computer

science. In: Rine, D.C. (ed.): Computer Science and Multiple-Valued logic, theory and applications.

North-Holland Publishing Company, 1977, p 87 - 107

Epstein, G./ Horn, A.: Finite limitations on a propositional calculus for affirmation and negation. Bulletin

Sect. Logic, Pol. Acad. Sci. Inst. Philos. Sociol., v 3, 1974, p 43-44

Feigenbaum, Edward A.: Soviet cybernetics and computer sciences. Communications of the ACM, v

4, n 12, December, 1961

Frieder, G./ Epstein, G./ Rine, W.: A Summary of the Development of Multiple-Valued Logic as related

to Computer Science. 1974 Symposium on Multiple-Valued Logic, 1974, p 310

Frieder, Gideon/ Luk. C.: Ternary Computer. in: WORKSHOP ON MICROPROGRAMMING, 5TH,

ANNUAL, PREPRINTS, 1972.: Elektronika, Sep 25- 26, 1972, 98p

Grosch, H. J. R.: Signed ternary arithmetic. Digital Computer Lab. Memo. M-1496, MIT, Cambridge,

Mass., May, 1952

Haberlin, H./ Müller, H.: Arithmetische Operationen mit binär codierten Ternärzahlen. Mitteilungen

AGEN, v 11, 1970, p 55-58

Hallworth, R.P./ Heath, F.G.: Semiconductor circuits for ternary logic. Institution of Electrical Engineers,

Monograph No. 482 E, Nov, 1961

– 20 –

Hanson, W. H.: Ternary Threshold Logic. IEEE -- Transactions on Electronic Computers, v EC-12, n 3,

June, 1963, p 191-197

Higuchi, Tatsuo/ Hoshi, Hisamitsu: Special-purpose ternary computer for digital filtering. Proceedings

of the eighth international symposium on multiple-valued logic 1978, Rosemont Illinois, 1978, p 47–

54

Ivaskiv, Yu. L.: Principles of multi-valued implementation schemes. Naukova Dumka, Kiev, 1971, (in

Russian)

Kaniel, A.: Trilogic, a three-level logic system provides greater memory density. EDN, Apr, 1973, p

80-83

Klimenko, Stanislav V.: Computer science in Russia: A personal view. IEEE Annals of the history of

computing, v 21, n 3, 1999

Knuth, D.: The Art of Computer Programming. Addison-Wesley, 1969

Lalanne, L.: , Comptes Rendus, v 11, Paris 1840, p 903 - 905

Lewin, M.H.: Retrieval of ordered lists from a content addressable memory. RCA Review, v 23, 1962,

p 215-229

Malinovski, B. N.: Istorija vychislitel’noj tekhniki v licakh. Kiev, 1995, (in Russian)

Merrill, R. D.: Tabular minimization procedure for ternary switching functions. IEEE Transactions on

Electronic Computers, v EC-15, n 4, Aug, 1966, p 578-585

Merrill, R. D.: Some Properties of Ternary Threshold Logic. IEEE Trans. on Electronic Computers, EC-

13, Oct., 1964, p 632-635.

Merrill, R. D.: Ternary logic in digital computers. Proceedings of the SHARE design automation project,

1965

Mine, H. et. al.: Four ternary arithmetic operations. Systems-Computer-Controls, v 2, 1971, pp. 46 –

54

Porat, D. I.: Three-valued digital systems. Institution of Electrical Engineers Proceedings, v 116, n 6,

June, 1969, p 947-954

Reader, T. D./ Quigley, R.J.: Research and development in fluid logic elements. UNIVAC monthly

press reports, nos. 1-8, NAS8-11021, July 1, 1963 – February 29, 1964

Rine, D. C.: An introduction to multiple valued logic. In: Rine, D.C. (ed.): Computer Science and Multiple-

Valued logic, theory and applications. North-Holland Publishing Company, 1977, p 3

Rine, D.C./ Epstein, G./ Frieder, G.: The development of multiple valued-logic as related to computer

science. In: Rine, D.C. (ed.): Computer Science and Multiple-Valued logic, theory and applications.

North-Holland Publishing Company, 1977, p 87-107

Robertson, James E.: In: Carr 1959

Rudins, George: Soviet Computers: A historical survey. In: Holland, Wade B. (ed.): Soviet Cybernetic

Review, v 4j n 1, RAND Corp, 1970

Rumjanzev, Dmitri: Down with the Bytes! An Interview with N.P. Brusenzov. Upgrade 33 (175),

August, 2004, (in Russian)

Santos, J./ Arango H./ Pascual, M.: A ternary storage Element using a conventional ferrit core. IEEE

Transactions on Electronic Computers, v EC-14, n 4, Apr, 1965, p 248

Schauer, R./ Steward, R./ Pohm, A./ Reid A.: Some applications of magnetic film parametron as logical

devices. IRE Trans. on Electronic Computers, v EC-19, 1960, p 315-320

Thelliez, S.: Introduction a l’etude des structures ternaires de commutation. Gordon & Breach, Paris,

1973, p 175-191

Varhshavsky, V.I., Bogolubov, I.N.: Ternary threshold elements and problems in their synthesis.

Transl. Proceedings, relay systems and finite automata, Burrough Corp., no. MT-63-257, article 39

– 21 –

Varhshavsky, V.I.: Ternary Majority Logic. Avtomatika i Telematika, v 25, n 5, 1964, p 673-684

Varshavsky, V.I./ Ovsievich, B.: Networks composed of ternary majority elements. IEEE Transactions

on Electronic Computers, v EC-14, n 5, Oct, 1965, p 730-733

Vranesic, Z.G./ Hamacher, H. C.: Multivalued versus binary high speed multipliers. 1971 Symp. Multiple-

Valued Design, Buffalo, May, 1971, p 42-53

Vranesic, Z.G./ Hamacher, H. C.: Threshold Logic in Fast Ternary Computers. Proceedings of the fifth

international symposium on multiple-valued logic, 1975, p 373-377

Ware, Willis. H. (ed.)./ Alexander, S.N./ Armer, P./ Astrahan, M.M./ Bers, L./ Goode, H.H./

Huskey, H.D./ Rubinoff, M.: Communications of the ACM, v 3, n 3, 1959, p 131-166; IRE Transactions

on Electronic Computer, v EC-9, n 1, Mar, 1960, p 72-120; Bulletin Provisional International

Computation Centre 10-11, July-Oct, 1960; RM-2541, RAND Corp., March 1, 1960

Yablonksij, S.V.: Functional Constructions in K-valued logic. Proceedings of Steklov Institute of

Mathematics, (in Russian), v 51, 1958

Yoeli, M./ Halpern, I.: Ternary arithmetic unit. Proceedings of the Institution of Electrical Engineers,

v 115, n 10, Oct, 1968, p 1385-1388

Yoeli, M./ Rinen, S.: Application of ternary algebra to the study of static hazards. Journal of ACM, v 9,

n 3, p 84-97, 1964

Zhogolev, Y. A.: The order code and an interpretative system for the Setun computer. USSR Comp.

Math. And Math. Physics (3), 1962, Oxford, Pergamon Press, p 563-578

– 22 –

Appendix A

List of Authors with more than 3 publications

Amount of Articles

(single and co-author)

Author Affiliation

7 Merrill, R. D. Electronic Sciences Lab, Lockheed Missiles & Space

Co., Palo Alto, Calif., USA

7 Vranesic, Z.G. University of Toronto, Ontario, CA

6 Frieder, Gideon State University of New York/ Buffalo, USA

5 Muzio, J.C. University, Winipeg, CA

5 Mouftah, H.T. Laval University, Quebec, CA

5 Rine, D. C. Department of Statistics and Computer Science, West

Virginia University, Morgantown, USA

4 Miller, D.M. University, Winipeg, CA

4 Arango, H. Department of Electrotecnia, Universidad Nacional del

Sur, Bahia Blanca, Buenos Aires, Argentina

4 Santos, J. Department of Electrotecnia, Universidad Nacional del

Sur, Bahia Blanca, Buenos Aires, Argentina

3 Hasegawa Kyoto University, Japan

3 Mine, H. Kyoto University, Japan

3 Higuchi, Tatsuo Department of Electronic Engineering, Tohoku University,

Aoba, Aramaki, Sendai, Japan

3 Thelliez, S. Universite de France Comte (?)

3 Yoeli, M. Faculty of Electrical Engineering, Technion, Haifa, Israel

3 Varhshavsky, V.I. Department of Mathematical Inst. of Academy of Science

of the USSR, Computer Center of Leningrad,

USSR

– 23 –

Appendix B

Examined documents on ternary logic, mathematics and computing

Author Title Publisher SETUN

mentioned

Abdul-Karim, M. AH.: A ternary J - K memory. Proceedings of the eighth international

symposium on multiple-

valued logic 1978, p 221–

225

no

Bate, J. A./ Muzio, J.

C.:

Three cell structures for

ternary cellular arrays.

Proceedings of the sixth international

symposium on multiple-

valued logic, 1976, Logan,

Utah, p 55–60

no

Frieder, Gideon/ Fong,

A./ Chao, C.y.:

A balanced ternary computer.

State University of New york/

Buffalo, 1973; Conference Record

of the 1973 International

Symposium on Multiple-Valued

Logic at Toronto, Canada, May,

1973, p 68-88

no

Frieder, Gideon/ Luk.

C.:

Ternary Computer. in: 5th Annual Workshop on

Microprogramming, 1972.:

Elektronika, Sep 25- 26, 1972;

State University of New york/

Buffalo, Report 32-72-MU,

SUNY/AB, 1972

yes

Frieder, Gideon/ Luk.

C.:

Algorithms for binary

coded balanced and ordinary

ternary operations.

IEEE Transactions on Computers,

v C-24, n 2, Feb, 1975, p

212-215

no

Hallworth, R.P./ Heath,

F.G.:

Semiconductor circuits

for ternary logic.

Institution of Electrical Engineers,

Monograph No. 482 E,

Nov 1961

yes

Hanson, W. H.: Ternary threshold logic. IEEE Transactions on Electronic

Computers, v EC-12, n 3, June,

1963, p 191-197

no

Higuchi, Tatsuo/

Kameyama, Michitaka:

Static-hazard-free T-gate

for ternary memory element

and its application

to ternary computers.

Proceedings of the international

symposium on multiple-valued

logic, 1976, p 127-134

no

Irving, Thurman A. Jr./

Shiva, Sajjan G./ Nagle,

H. Troy Jr.:

Flip - Flops for multiplevalued

logic.

IEEE Transactions on Computers,

v C-25, n 2, Mar, 1976, p

237-246

no

Kooi, C.F./ Weaver,

J.L.:

Non-Linear Hall Effect

Ternary Logic Element.

Solid State Electronics, Pergamon

Press, v 7, n.5, 1964, p

311-321

no

– 24 –

Lewin, M.H.: Retrieval of ordered lists

from a content addressable

memory.

RCA Review, v 23, n 2, 1962, p

215-229

no

Merrill, R. D.: Some Properties of Ternary

Threshold Logic.

IEEE Trans. on Electronic Computers,

v EC-13, Oct. 1964, p

632-635.

no

Merrill, R. D.: Ternary logic in digital

computers.

Proceedings of the SHARE design

automation project, Lockhead

Missiles & Space Co., Palo

Alto, Calif., Apr. 19, 1965

yes

Merrill, R.D.: A tabular minimization

procedure for ternary

switching functions.

IEEE Transactions on Electronic

Computers, v EC-15, n 4, Aug,

1966, p 578-585

yes

Merrill, R.D.: Symmetric ternary

switching functions –

Their detection and realization

with threshold

logic.

IEEE Transactions on Electronic

Computers, v EC-16, n 5, Oct,

1967, p 624-637

no

Miller, D.M./ Muzio,

J.C.:

Decomposition of ternary

switching functions.

Proceedings of 1973 International

Symposium on Multiple-

Valued Logic, 1973 p 156-165

no

Miller, D.M./ Muzio,

J.C.:

A ternary universal decision

element.

Notre Dame Journal of Formal

Logic, v 17, no 4, 1976, p 632-

637

no

Mine, H. / Hasegawa,

T./ Ikeda, M./ Shintari,

T.

Tri-stable circuit using

FET.

Electronics and Communications

in Japan, v 51, n 11, Nov,

1968, p 147-149

no

Morris, D.J/ Alexander,

W.:

An introduction to the

ternary code number system.

Electronic Enigneering, v 32,

1960, p 554

no

Mouftah, H. T., Jordan,

I.B.:

Integrated circuits for

ternary logic.

Proceedings of the 1974 international

symposium on multivalued

logic, 1974

p 285-302

no

Mouftah, H. T., Jordan,

I.B.:

A design technique for an

integrable ternary arithmetic

unit.

Proceedings of the fith international

symposium on multiplevalued

logic, 1975, p 359-372

no

Mouftah, H. T./ Smith,

K. C./ Vranesic, Z. G.:

Ternary logic in a

positional control system.

Proceedings of the international

symposium on multiple-valued

logic, 1976, p 135-141

no

Mouftah, H. T.: A study on the implementation

of three-valued

logic.

Proceedings of the sixth international

symposium on multiple-

valued logic, 1976, Logan,

Utah, p 55–60

no

– 25 –

Porat, D. I.: Three-valued digital systems.

Institution of Electrical Engineers

Proceedings, v 116, n 6,

June, 1969, p 947-954

yes

Rath, Shri Sudarsan: A ternary Flip - Flop circuit.

International Journal of Electronics,

v 38, n 1, Jan, 1975, p

41-47

yes

Reid, J. B.: Letter to the Editor [Existence

of trinary calculation

logic].

Communications of the ACM, v

3, n 3, 1960

no

Rescher, N.: Many-Valued Logic. McGraw-Hill, New York, 1969 no

Rine, D.C./ Frieder,

G./ Epstein, G.:

Summary of the development

of multiplevalued

logic as related to

computer science.

Proceedings of the 1974 international

symposium on multivalued

logic, 1974, p 303-314

yes

Rine, D.C./ Frieder,

G./ Epstein, G.:

Computer Science and

Multiple-Valued logic,

theory and applications.

In: Rine, D.C. (ed.): North-

Holland Publishing Company,

1977

yes

Rine, D.C.: Basic concepts of multiple-

discrete valued logic

and decision theory.

Information and Control, v 22,

n 4, May, 1973, p 320-352

no

Rosser, J. B./ Turquette,

A. R.:

Many-Valued Logics. Amsterdam, North-Holland,

1952

no

Santos, J./ Arango H.,

Lorenzo, F.:

Threshold synthesis of

ternary digital systems.

IEEE Transactions on Electronic

Computers, v EC-15, n 4, Feb

1966, p 105-107

no

Santos, J./ Arango H./

Pascual, M.:

A ternary storage Element

using a conventional

ferrit core.

IEEE Transactions on Electronic

Computers, v EC-14, n 4, Apr,

1965, p 248

no

Santos, J./ Arango H.: On the analysis of threevalued

digital systems.

Proceedings of the 1964 AFIPS

Conference, v 25, 1964, p 463-

475

no

Squires, Stephen L.: Microprogrammed Operations

for a Three-

Valued Logic Simulator.

Proceedings on the Workshop

on Microprogramming, 1974, p

262-266

no

Varshavsky, V./

Ovsievich, B.:

Networks composed of

ternary majority elements.

IEEE Transactions on Electronic

Computers, v EC-14, n 5, Oct,

1965, p 730-733

no

Vranesic, Z. G./ Smith,

K. C.:

Engineering aspects of

multi-valued logic systems.

IEEE Computer, v 7, 1974, p

34-41

yes

– 26 –

Vranesic, Z.G./

Druseta, A./ Sedra,

A.S.:

Application of multi

threshold elements in the

realisation of many valued

logic networks.

IEEE Transactions on Computers,

v C-23, n 11, Nov, 1974, p

1194-1198

no

Vranesic, Z.G./ Hamacher,

H. C.:

Multivalued versus binary

high speed multipliers.

1971 Symposium on Multiple-

Valued Design, Buffalo, May,

1971, p 42-53

no

Vranesic, Z.G./ Hamacher,

H. C.:

Threshold Logic in Fast

Ternary Computers.

Proceedings of the fith international

symposium on multiplevalued

logic, 1975, p 373-377

yes

Yoeli , M./ Rosenfeld,

G.:

Logical design of ternary

switching circuits.

IEEE Transactions on Electronic

Computers, v EC-14, n 1, Feb,

1965, p 19-29

no

Yoeli, M./ Halpern, I.: Ternary arithmetic unit. Proceedings of the Institution of

Electrical Engineers, v 115, n

10, Oct, 1968, p 1385-1388

yes

Yoeli, M./ Rinon, S.: Application of ternary

algebra to the study of

static hazards.

Journal of the ACM, v 11, n 1,

Jan, 1964, p 84-97

no

Zinovev, A. A.: Philosophical problems of

many-valued logic.

D. Reidel Publishing,

Dordrecht-Holland, 1963

no