The Committee noted that there were two obligations on Dr Howlett: Minute 3(i) - assessment of the value of the Data Products disc, and Minute 3(ii) - critical report on the multi-access system. Apart from this no points were raised.
The Committee noted with approval the continued high standard of reliability of the machine and the continued increase in the total throughput of work. It was clear that the multi-access system was not degrading the performance of Atlas but in fact improving it, by using profitably some of the idle time.
As this was by far the most important item on the agenda, it was taken first, and discussion on this subject occupied the greater part of the meeting.
Dr Howlett had written to all members as follows:
This letter had been sent out on Friday 11th July, but as a consequence of postal delays most of the members had not received it by the time of the meeting.
After Members had read Dr Howlett's letter and paper ACC/69/15, which was tabled, the Chairman introduced a general discussion of the new situation by saying that the decision of ICL to abandon the 1908A nullified one of the aims of the plans for computer purchase formulated by the Laboratory and the Computer Board: to give compatibility over a wide area by installing 1906As in some six to eight universities, with a guarantee of top cover provided by 1908As at Chilton and Manchester and the possibility of later extension to 1908A at selected universities. Mr Rutterford said that the Computer Board would need to reconsider its plans and would do so at its next meeting on 23rd July. Mr Llewelyn had similar plans for a compatible network of 1900-range machines for Min Tech's own establishments, including 1906As and 1908As; it was likely that they would continue with these as far as the 1906As. The Committee noted that the only alternatives at present to the 1907A were the 360/85 and the CDC 7600 (both of which were now built) and that a decision between proposing the buying of one of these machines and waiting for ICL's new plans to mature involved balancing the needs for provision of computer power with the need to support the British computer industry.
Mr Hall joined the meeting at this point (Mr Humphreys had been delayed by a traffic accident). He outlined the reasons which had led ICL to decide to abandon the 1908A. They could not continue to maintain two incompatible ranges (ICT 1900 and EE System 4) and had mounted an intensive and detailed study of a number of possible architectures for a new range. This study had been completed only recently. It gave an unequivocal choice of architecture (which was in fact a development of Professor Kilburn's design for his new machine at Manchester) and strong arguments for embarking very quickly on the design of a machine near the top of the new range, comparable in power to the 1908A. By starting near the top one avoided the problems one met in trying to extend upwards a design which had been formulated originally with the needs of medium-power machines in mind. A machine of the new design of 1908A power would be much simpler and physically smaller than the 1908A itself; and the architecture would permit machines of much greater power still to be built, whereas they saw no way to extend the 1900 series beyond the 1908A. Their Board had accepted these arguments and decided not to continue with the 1908A, because this would absorb technical effort which would be better used in the new range and result in a machine which would be overtaken very shortly by the new range. ICL were very conscious of the compatibility problems facing their customers and would produce Bridgeware (both hardware and software) to enable them to transfer programs from 1900s and System4s to the new machine. They themselves would have the difficult problem to solve of finding how to introduce the new range without damaging their own market.
Because it had not been possible to arrange this meeting until after the last meeting of the UST Board in this session, this paper had been submitted to the Board (paper USTB-262) before the Committee had seen it. The Board had approved these works, subject to the approval of the Committee. The Committee approved.
This paper also has been seen by the UST Board, who had given conditional approval similarly. The Committee recognised that the proposed equipment would be a powerful addition to the Laboratory's facilities (noting that users would need to come to the Laboratory) but expressed some doubts about the value of the work to which users would put it. Dr Howlett emphasized the educational aspect - because of the Laboratory's central position it would be able to gather ideas and experience concerning the processing of visual information from research workers all over the country, and to spread this knowledge widely. Professor Barnard confirmed that in the particular case of statistics there was a strong need to study the use of visual representation. The Committee finally expressed itself satisfied and gave its approval of the purchase of the DEC system at a cost not exceeding £65,000.
This was for information. The Laboratory had been asked by the Chairman of the Council to support this work, the UST Board had approved it and finance Division had promised to provide the money needed.
It was agreed that the date of a meeting in October or November should be settled by correspondence.
This was the last occasion on which Lord Halsbury would chair the Atlas Committee meetings, as his term of office ended on 30th September. He would be succeeded by Sir Rudolf Peirls. Dr Howlett expressed the Laboratory's thanks to Lord Halsbury for the great amount of work he had done in guiding its progress over the last four years, and his personal thanks for the deep personal interest Lord Halsbury had taken both in the Laboratory's work and in the people working there. The Committee warmly concurred.
