Bull History FAQ
this FAQ (Frequently Asked Questions) wish to answer miscellaneous and informal -sometimes impertinent- questions related to the history of the company and its relations with the rest of the computing world. The following are examples of those Q&A.
Q: What the word Bull means? what are the relations with the animal?
A: Bull is really a person name, the name of a Norwegian engineer who designed a punched card data processing machine for its insurance company. Frederik Bull filed patents that were at the origin of Compagnie des Machines Bull.
Q: Why naming a French company after a Norwegian ?
A: Frederik Bull's patents were acquired by a Swiss company Egli AG. But, the technical heir of Frederik Bull, Kurt Andreas Knutsen preferred to settle in Paris in 1930.
Q: Did Frederik Bull really invented the tabulator?
A: No. The punched card processing has been developed by Hermann Hollerith in 1884. Hermann Hollerith was at the origin of IBM Business Machines. Frederik Bull first patents were related to the preselection of data inside punched cards reader in 1919 (35 years after Hollerith).
Q: What were the reasons for the bad relations between the Callies family and De Gaulle's government?
A: The main shareholders of Compagnie des Machines Bull were the Callies family
relative of Michelin family. They were secretive and very reluctant to any government
intervention in their business. So, they essentially ignore the needs of the military
agencies in the 1950s and let that market to SEA.
In addition, Bull had continued its activities in 1940-1944 under Vichy regime and German
occupation, and although several actions inside the Company were directed against Germans,
they were, of course, not made under Gaullist allegiance. Later, a significant part of CMB
management was coming from French Navy's officers who have been sacked as hostile to
Gaullists.
So, when Joseph Callies faced a financial crisis in the 1960s, he had less credits to
request favors for the government of Charles De Gaulle. .
Q: What was the policy of CII and Bull vis à vis the Comecon countries?
A: While having closer relations with Eastern Europe countries than the American manufacturers, CII and Bull have obeyed the rules established by COCOM in their sales to USSR and other Comecon countries.
Q: Why Compagnie des Machines Bull did not succeed in making an agreement with the late F-H Raymond, founder and CEO of the SEA ?
SEA was a start-up that entered the electronic computer field (analog then digital)
almost as soon as the US industry did and at a time where Compagnie des Machines Bull was
in the electro-mechanical business. Competition ignited when Bull entered the general
computer market with its drum computer gamma ET) and SEA started to develop business
tape computers.
However, there was an agreement signed in the early 1960s based on Bull would
have marketed products designed by SEA (at that time the personal computer CAB500 - a
cheaper IBM 1620- and the medium size tape business computer 3900). Bull sold only a
handful of CAB500 and not any 3900, far from the agreement expectations. So, Bull was sued
by SEA in the court and was forced to pay compensations. Bull had commitments to RCA by
selling the Gamma 30 (in lieu of SEA 3900) and their sales network did not show a
significant interest in scientific computers.
Raymond was considering that Machines Bull engineers were stubbornly devoted to the
punched card business and CMB had to steal engineers from the SEA's staff for their Gamma
3 and Gamma 60 electronic computers
.So, when Bull fell into GE's hands, Raymond was among the first to ask the French
government to ignore what he called Bull-GECO. However, Raymond was put aside by the Plan
Calcul and became professor at CNAM. While, he had still a resentment against many
managers of CII and Bull, he became a friend to our association (FEB) and he was always
glad to explain his views about how the French computer industry would have been. Those
views remained pertinent.
Q: Why CII-Honeywell-Bull got nationalized in 1982?
A: There is an obvious, but very simplified, answer. A leftist coalition including
Communists took over the government, after the election of François Mitterrand in may
1981. In fact, France had a centuries-old tradition of State owned enterprises. It started
by "Manufactures d'Etat" in the XVIIth century, but nationalization became quite
frequent when companies created by capitalistic entrepreneurs started to fledge in the
1930s. French railways, French electricity, important bank and insurance companies were
nationalized under right and left aisles governments. In addition to those sectors that
could be considered as potential monopolies, some industrials got nationalized either as
an economic sanction about their war activities, as Renault, or because the industry was
considered as strategic and was menaced by a foreign take-over or a disappearance, like
oil production that was concentrated into nationalized Elf-Aquitaine and the steel
industry that concentrated in Usinor in the 1960s.
The first nomination for nationalizing Bull was made by the Communist Union when the
financial state of Compagnie des Machines Bull came in the public knowledge in 1964. The
right wing government allowed the take-over by GE and help the private sector (Thomson-CSF
and CGE) to build a heavily state subsidized CII. After that, the liberal government of
Giscard d'Estaing allowed a take-over of CII by Honeywell-Bull. The new company CII-HB was
in majority governed by French CGE and later by Saint-Gobain. The Left added CII-HB on a
list of companies to be nationalized because they were state sponsored and not responsible
before their main sponsor. The "Programme Commun de Gouvernement" also included
the major French industrial companies that were employing hundred of thousands of workers,
were unable to get financing from the market and were often menaced of decline. The list
included Saint-Gobain and Thomson.
After the success of the Left in 1981, CII-Honeywell-Bull was in conflict with its main
shareholders about an attempted take-over of Olivetti by Saint-Gobain and it entered
financial difficulties. So, Jean-Pierre Brulé was dismissed of his chairman's position,
with the agreement of Honeywell , and the State take-over negotiation started. Private
shareholders, Honeywell being the largest one, got cash in exchanges of their share. Then,
the government started to reorganize the industry by merging into Bull the activities of
other nationalized companies (essentially Thomson and CGE) and made Groupe Bull an
independent state-owned company, a statute it kept until 1994.
At the beginning, there was a trend into the government, and among some of its engineers
and workers, to consider Bull as a defense "arsenal", that forgot all
considerations of profit and cost to achieve technical goals.. In fact, that tendency
faded rapidly.
Q: With what American Companies Groupe Bull has cooperated?
A: In the 1930s, Bull's shareholder Emile Genon attempted a merger with Remington-Rand
that failed in front of the hostility of French shareholders of HW
Egli Bull. A cross-license and sales agreement with Remington-Rand was concluded in 1949
and RR has been a distributor of Compagnie des Machines Bull in USA. In 1961, Bull
concluded a distribution agreement with RCA as a reseller and
manufacturer under license of RCA 301. In 1964, General Electric acquired
the majority of Bull. When GE walked out from the computer business in 1970, Honeywell
took over its interests in Bull. In the 1980s, Bull negotiate some licenses agreements
with Ridge, MIPS, Storagetek and finally IBM.
Among the French companies that merge into Groupe Bull, CAE was Ramo-Wooldridge
then Scientific Data Systems licensee, Thomson was
a distributor and a shareholder of Fortune, Transac had acquired
the Convergent Technology CTOS products license that was carry
on Bull.
Q: What genuine French computers products have been sold in quantity in the USA?
A: Punched cards equipment by Remington Rand in the 1950s. GE-58 then Level-61 were distributed in the 1965-1975 period by GE and Honeywell. Level-64 and DPS-7-x5 were sold by Honeywell from 1974 to 1980s -several hundreds of medium-size computers-. Line and page printers designed and produced in Belfort have been sold on GCOS8 and on IBM mainframes.
Q: Was Honeywell-Bull Distributed System Architecture derived from Cyclades INRIA project ?
A: While Cyclades was developed in Roquencourt, at a walking distance from Louveciennes's CII lab, the Cyclades experimental project initiated by Louis Pouzin did not evolve into CII's NNA -the predecessor of DSA. CII designers were perfectly aware of Cyclades design, but they diverge from the Cyclades options that were rejected by standardization bodies in CCITT. PTTs and notably French PTT preferred the so-called "connection" protocols -closer to the Telex culture- to the "connection-less" datagram protocol that Pouzin had borrowed from Arpanet. After the fact, it may seem unhappy that this architecture choice made Honeywell moved away from what was the direction of Internet future standard. But, before deregulation, the PTT choices were unavoidable and DSA was following them. Note that PTT imposed the X-25 standard because it mapped well on the then sacrosanct Telex that emphasized end-to-end transmission control.
Q: Why Bull adopted Unix proselytism in the 1980s?
A: At a time where Bull management and its government shareholder was essentially
concerned in the long term future, Jacques Stern identified the market irresistible
trend towards open systems. He identified that the dominant IBM business model was based
on having customers captive by IBM standards, IBM architecture and IBM proprietary
software. Neglecting somewhat the dynamism of new entrants, he imagined to reposition Bull
as the alternative to the IBM model by forcing the formal standardization of UNIX and
phasing out the Bull proprietary systems into open systems.
Unhappily, Bull losed time and money in building a UNIX alternative from many disparate
parts. The market response to open standards remained to be very conservative and to
associate too often open standards and downsizing to the Wintel world. Anyway, it is
unlikely that Bull could have maintained four separate lines of GCOS computers, at a time
where decreasing hardware costs leading to decreasing prices were deeply cutting sales
margins.
Q: What was the Bull's management attitude towards Multics?
A: Bull-General Electric was very enthusiastic when Multics was initially announced in 1965. But the long infancy of this operating system decreased the interest of initial prospects. The 1966 crisis of the whole GE-600 line reinforced the Bull's management feeling that the most profitable part of the business was the world of business applications where customers were committing their company for a minimum of 10 years of fruitful contracts. Specially after Honeywell took over, the GCOS product line was focused towards this bread and butter target. Multics remained an experimental system done by and for scientists. When French educational and engineering accounts were brought by CII merger, then Multics started to enter the Bull's market and many orders were secured. But Phoenix failed to produce a Multics version of the bad famed DPS88 system and NEC processors took over. NEC had not been licensed to Multics, probably because it had been financed by DARPA money. The changes to the DPS-90 to run MULTICS were, in Japanese parlance, "very difficult". The idea of running Multics on a smaller machine, such as level 6, does not seem to have attracted interest in Europe. The "Flower" project was more seen as a "struggle for life" defensive by Boston engineering than a serious business case, specially because Bull's management has already made its mind in favor of a standard UNIX.
Q: What was specific to the European and the French markets from the technical point of view?
A: Two countries influenced seriously the European markets and Bull was strong in those
two: Sweden and France. Sweden was the first to prescribe noise standards for operators
staying in the computer room and the first to edict rules on file privacy. France was
edicting laws for the protection of the French language.
In addition, the major difference between USA was the power supply problems where USA run
on 60Hz while Europe adopted 50Hz. there was also a voltage difference (110v vs.
220v).
Such differences were important in the world of electromechanical equipments and was
resolved by the Japanese who had the same differences between Osaka and Tokyo.
Q: Bull encountered financial difficulties in 1962-1964, in 1979-1982, in 1992-95. Was it for the same reasons?
A: The financial crises had common aspects as some common consequences (restructurations, lay-offs...). In the industry in general, about on the same time, computer companies face the same general problems and many of them merge or disappear as a consequence of them. Although, with important changes in shareholders and management, Bull essentially succeed to stay afloat.
The 1962-1964 was a growth crisis, where Bull shareholders were not able to finance the
leased park growth, destabilized by the fast renewal of electronic computers. In addition,
the Bull's market was to small to absorb the cost of the development of Gamma 60. And
healthy economics was still absent from the Bull's culture.
In 1979-1982, the economics condition of the industry were dramatically changed by the IBM unbundling and the shift away off the leased products. It was marked by an explosion of sales and a certain disruption of the manufacturing. In addition, the change caught Bull in a technological transition that caused new product delays.
The most recent 1990s crisis was the consequence of the decline of hardware prices and the arrival of open systems that stopped the growth of proprietary products. A company like Bull that had distributed the responsibility of its product line all across the planet. It had also many sources of overhead can had to be drastically cut out. But the European culture and laws required a long time to reorganize the company.
Q: How the company choose to offer IBM incompatible mainframes?
A: In fact, the strategy was designed well before IBM S/360 invention. Compagnie des
Machines Bull history have seen several patents issues related to the punched cards
mechanism and encoding and had to differentiate as much from IBM as its customers would
allow.
As for the GE side, there was initially the idea that the excellence of the products
should be sufficient to grasp customers from their original supplier. The Honeywell
management, on the contrary, were attempting to build the strategy with enough hooks to
catch customers, but enough differentiation to be independent. H-200, Level 64 and Level
62 were founded on that basis: keep data compatibility with IBM, allow the conversion of
programs (if possible one-way).
The original "plan calcul" strategy was not very different from Honeywell's. But
Unidata's strategy had to align on Siemens that , following RCA and Japanese Hitachi, was
tempted by a "almost clone" strategy. In fact, IBM customers love clones,
because they give them bargaining power and alternatives to the sometimes predator
attitude of IBM salesmen. But, the clone strategy required a technology supremacy over
IBM, the architecture advances of Amdahl does not suffice. CII, nor Honeywell and Bull did
not have the resources to achieve a permanent technology supremacy over IBM, even if Bull
introduced a CMOS mainframe system before IBM.
The data incompatibility of GCOS8 systems with IBM has, in addition, historical
reasons. The line was born before 1964 and received limited changes. As a customer said
once, there is only a single thing more difficult than converting from S/370 to DPS-8,
it's to go back.
Q: Did ever Bull or Honeywell consider to enter the IBM mainframe clone market?
A: In 1981, CII-HB was approached by Gene Amdahl who was looking for investors for its
wafer-scale integration venture Trilogy Inc that was planning to sell high performances
IBM compatible mainframes. CII-HB accepted, as did Digital, essentially for having access
to that technology for future DPS-7. Some consideration was made of reselling the Trilogy
machine for the European market. But actually, Trilogy failed before Bull went longer than
a marketing investigation.
NEC had produced IBM compatible DIPS machines for the internal market of NTT (Japanese
telephone) as an alternate supplier to Fujitsu and Hitachi. Those machines were built on
the same technology as ACOS systems. But NEC never envisioned to sell them in the open
market, even in Japan, and never proposed them to Bull.
Q: What computer Level 64 architecture was inspired from: IBM FS, IBM 370, Honeywell NSA , Multics, Mu-5, Intel 286, Burroughs B-5000,...?
A: Actually many of those Complex instruction sets were unknown to the architects. The architectures that were known and have influenced somewhat the level 64 were: IBM S/360, GE-645 and Burroughs B-5000.
Honeywell Level 66 was designed after Level 64 and the architecture had slowly penetrate GCOS-8 operating system. Intel 286 was not born and many issues Intel encountered had been faced many years before by Honeywell Bull designers (64KB segments, mode switching etc...). The IBM FS development was contemporary, but its rumored features did not impress very much the Honeywell and Bull architects, probably because their culture dismissed a rupture of compatibility FS was envisioning. Manchester University work on a compiler oriented system (MU-5) and influence of it in ICL 2900, was rumored but it only confirmed that the language concern of ex-Burroughs, now GE, was real.
Data compatibility with S/360 was part of the specifications and that lead to the
32-bits words, EBCDIC 8-bits data and format for decimal and floating point.
Segmented address space and the ring protection mechanism were borrowed from Multics.
There has been different schools for the instruction definition:
The result was trying to satisfy all of them. The "interior decor",
interpreted by microprograms on a model dependent hardware structure, offered a basic
register architecture supplemented by a stack mechanism designed for the efficient
processing of block-structured languages dynamic variables addressed from nested
procedures, indirect addressing was available and allowed some descriptors capabilities
with data protection and a "micro kernel" in firmware implemented multi-tasking
primitives directly coming from Dijkstar's THE operating system.
Those compromise were not made in a hurry and after the use of those features has been
validated by software people, so nobody complained seriously after them during the thirty
years of life of that architecture. NEC had the architecture implemented in processors
that were in the late 1980s more powerful than IBM systems.
Q: When Honeywell did actually plan to abandon computers?
A: It has been said in France that such a plan existed already since 1974 and that the
acceptance of the loss of the majority in CII-Honeywell-Bull in exchange of the absorption
of CII and advantageous conditions in case of CII-HB nationalization was due to such a
plan.
Each financially conscious company being in distinct business do reexamine its commitments
and reposition its investments at least yearly. The enthusiasm of the Honeywell
shareholders for computers had certainly decreased since 1967-1970, and some members
of the Board may have mentioned a bail-out hypothesis. Possibly also the guarantees that
Honeywell asked and obtained from the French government in 1975 in case of changes in
French shareholders (read in case of nationalization) may have been interpreted as signal
of a potential withdrawal.
But it is not before 1979 that signs of abnormal worry transpired at the operational
level. In 1979, Honeywell recognized that there was no more hope to have
customers switching from IBM to HIS .DPS-8. The GCOS-8 business was
rescued in 1984 with the import of NEC compatible systems. It is probably at the same
moment that a decision to sell the business was taken. The solution that was the less
disturbing to the customers was a take-over by a mixture of NEC and Bull, but other
solutions may have been analyzed..
Finally, NEC suggested to take a minority share, apparently, because they were unable to
staff the HIS management with strong Japanese managers, a condition thought necessary for
a majority take-over.
While HIS was only a sales outlet to NEC, it was an essential supplier for Bull. A
take-over of HIS by a computer competitor would have lead to a quick disappearance of
GCOS8..
Q: What are the origins of Bull-NEC relationship?
A: The relations between NEC and Bull start at the time of the take-over by Honeywell
of the General Electric business. Compagnie des Machines Bull had Mitsubishi Office as
distributor in Japan, that imported one Gamma 60. General Electric had Toshiba that was a
licensee of GE for a variety of products including computers. Honeywell had Yamatake for
control systems and had licensed Nippon Electric for computers. This set of license
agreements concluded by Japanese companies had been engineered at the end of the 50s by
the MITI.
When the industry shake-out occurred around 1970, MITI "recommended" also merges
in the Japanese industry and Toshiba transferred its mainframe computer business to NEC.
NEC negotiated with Honeywell its license agreements that cover the whole set of computers
including the Honeywell-Bull Level-64.
When CII-HB was created in 1975, the new products of that company did not belong to the
scope of agreements (e.g. network architecture, DPS-7...). The control of the
Honeywell-NEC belonged to Honeywell, although CII-HB's people pay regular visit to Japan
and was honoring the previous agreements.
It was in 1984, that Honeywell first and Bull subsequently observed that their top of the
line projects for GCOS-8 (and also GCOS7) were in jeopardy and that buying back some
products from NEC could close the gap in their product lines. The NEC sales in USA and
Europe and the related technical exchanges were an incentive for NEC to take a minority
participation in the capital of both companies, that finally was converted into 17% of
Bull shares..
Q: What were the reasons of the acquisition of Zenith by Bull?
A: After the Honeywell acquisition, Bull was committed to the worldwide market. Bull was convinced that there was no way to attack successfully the US PC market from its French base, that was trailing behind IBM and Compaq. Zenith was publicly for sale and had a competitive engineering and sales organization. The initial price asked for Zenith was challenged by its shareholders and the "goodwill" that Bull finally agreed to pay was probably too high, because it was not obvious that a foreign company could durably keep the DoD contracts and because the engineering assets were very volatile.
Q: Why the acquisition of R2E and the Micral did not give a strong position in the PC market?
A: R2E and Bull-Micral did not identify at all the domestic market for the PC and they position the product as a professional tool only, contrarily to Apple or the ill-fated Thomson TO7. Their product faced an internal competition with word-processing TTX and with Questar 400 on the same professional market. They differ essentially by the distribution channel that used essentially independent retailers.
Q: What is CMC7?
A: CMC-7 is a encoding of characters for check recognition. Coming after the E13B American standard, it allow the coding of alphabetic and numeric on a kind of Bar Code painted over a human recognizable font. CMS7 was adopted by French banks in the late 1950s and is still used today. Characters were printed with an ink including magnetic particles.
Q: What were the relationship between Roland Moreno "the inventor" of smart card and Bull?
<to be answered>
Q: What have been the relations of CII and Bull with the British manufacturer ICL?
A: Compagnie des Machines Bull was an OEM supplier to BTM, a predecessor of ICT, future
core of ICL.
CII had attempted to create links with ICL. They look together at the possibilities to
harmonize their product lines, but ICL was not ready to yield on the design of the 2900
architecture. So, ICL and CII content themselves with establishing a common entity,
Multinational Data, with Control Data to promote common standards . No great outputs
came
from that entity and CII teamed with Siemens, the product line of which was apparently
easier to integrate.
After the CII-HB creation, and before the take-over of Fujitsu on ICL, ICL approached Bull
in attempting to promote its data base machine.
Q: Is it true that CI-Honeywell-Bull was involved with USA's Department of Defense in ADA design? Had Bull used ADA as an implementation language for some products?
A: Yes, a language research team in CII, lead by Jean Ichbiah had written specifications
and started prototyping an implementation language called LIS (System
Implementation Language) when the DoD issued its first Strawman RFP for a language that
should be used all across the US Services. After, the Honeywell merger, Jean Ichbiah made
its proposal backed by Honeywell Federal Division, using Multics, as the Green Team
proposal... and , after several rounds of refinement, won the ADA RFP.
CII-HB had contemplated the use of ADA, but its main product GCOS 7 was already using a
high level implementation language. The French military had selected a different language
LTR for their use, before ADA was created. So, ADA, while originated in Bull, was not used
internally. Jean Ichbiah and its team eventually founded their company Alsys, bought later
by Thomson-CSF.
Q: How do you compare NEC and Honeywell product lines?
A: NEC developed four product lines from Honeywell licenses: Minicomputers (from Level 6), ACOS2 (from Level 62), ACOS 4 (from Level 64) and ACOS 6 (after their take-over of Toshiba main frames, from Level 66). In addition, they produce DIPS as a special for NTT and SX a supercomputer line. They also develop several lines of personal computers that were taking the major part of the Japanese market (PC9800 being the professional market product line running a special version of MS-DOS and later Windows).