|An appeal for more information|
On 27 January 2000 Charles Norrie sent me email from London. He raised two questions in which I have a particular interest.
Firstly he asked whether I had any film or video of working Julius ( electro-mechanical ) totalisators. I lament that I have not seen a Julius totalisator in action and have often wondered about the existence of film footage. I have included Charles' email here as an appeal to anyone who may be able to provide more information on these systems film or otherwise.
Secondly Charles is interested in the relationship between these Julius totalisators and computers. I have often contemplated this question. It is curious to think of a multi user real time system with distributed points of sale in 1913. The largest of these systems in operation, that I am aware of, was in Longchamps in 1928 with 273 terminals. (Postscript - I have recently read a post world war 2 booklet titled The Julius Premier Totemobile, which lists all installations by number of terminals. It shows White City Stadium London as having 320 terminals.)
A Julius tote system was built and tested in Sydney capable of supporting 1000 terminals and sales in excess of 4000 per second. Did mechanical computers exist? If not, had these systems been more widely known would they have been classified as mechanical computers?
There are many analogies with computers and the metrics are very similar e.g. number of terminals, sales per second, time to print ticket and error rate.
In the Photo Gallery Continued Chapter of this website, the page loaded by selecting the image thumbnail under the heading Figures from George Julius' paper presented to the Institution of Engineers Australia in 1920, with associated text starting FIG 9 from George Julius' paper ... contains items I have annotated in George Julius' paper putting it into modern day terminology. Putting them all together George is describing a Large Scale, Low Response Time, Real Time, Networked, Reliable, Multi User System in 1920!.
These systems had input devices, the Ticket Issuing Machines and other configuration devices. They also had output devices, Total Counters, Odds Indicators and Ticket Printers. A mechanical device was used for adjusting the amount of commission deducted from the pool.
An electromechanical device called a Scanner or Distributor is a Time Division Multiplexer which multiplexed the impulses from up to 16 Ticket Issuing Machines onto one of the many solenoids in a multitude of electromechanical shaft adders in one of these systems.
Electromechanical devices called a Storage Screws, in computer terms would be called a buffer memory. Another analogy in the design of these storage screws relates to the electronics concept of servo systems. There is an angular velocity control system associated with this device. In the paper George presented to the Institution Of Engineers Australia in 1920, he refers to this control system as a variable speed friction gear and its application reminds me of a closed loop servo system.
Another digital computer analogy with these electromechanical systems is parallel processing. The Julius Tote Engineers used to remind me that the computer systems I introduced to replace Julius Totes could not do something the Julius totes could. The Julius totes by virtue of the epicyclic gear train in the adders, could simultaneously record multiple transactions. My computer systems, although being loosely coupled multiprocessor systems, were not capable of parallel processing and consequently had to process all transactions sequentially.
I have had a lot to do with sequencers in Electronics and Computer Engineering. After a couple of days spent in the Totalisator Computer Room at a racetrack in Kota Kinabalu, analysing a fault with a PDP11/44 minicomputer based transaction processor, I finally located the faulty component, an IC (Integrated Circuit) implementing a BCD (Binary Coded Decimal) counter in one of the sequencers in the CPU (Central Processing Unit). This fault was amongst the most difficult of faults to isolate I have worked on and consequently is amongst the faults that took the maximum time I have spent on isolating a single fault. I have a J8, a Julius Tote TIM (Ticket Issuing Machine) in my collection of memorabilia. This like other models has quite an elaborate mechanical sequencer in it implemented by a cam shaft.
We would be very interested in hearing some opinions on this matter. We can be contacted via email@example.com The system at Harringay was the last Julius tote to cease operation. It operated until 1987.
Postscript - It is now June 2005 and I have just received mail from Caracas. I was informed that the Julius tote there is still in operation. After 48 years of operation I have been asked if I have any information on how to make adjustments to the system to bring it up-to-date!
I asked Max Burnet, Vice President of the Australian Computer Museum Society and ex CEO of DEC Australia, about mechanical computers. He replied:
The catagory of mechanical computers never got established. Relay and valves yes, but mechanical no. The ATL tote certainly deserved to be classified as a mechanical computer.
I have viewed your site on George Alfred Julius and the automatic totalisator with great interest. May I congratulate you on it.
There are several purposes in sending this e-mail to (i) to introduce myself as someone who has some considerable interest in Julius (ii) to write about the Harringay tote which GLIAS (the Greater London Industrial Archaeology Society) recorded working during its last days and (iii) to ask you whether you know of any film or video sources of the operation of Julius machines.
I became interested in the Harringay tote machine in 1987 - I had not heard of Julius before that - and particularly in its operation as a kind of mechanical (or electro-mechanical) computer. The machine at Harringay must have been much as it was at the time of its major update in the 1930’s - though drums of an earlier installation were still in place. The only major modification was the addition of magnetic counters which transferred running counts to a computer which generated a video presentation of the running odds on the win bet. The odds calculating machinery had been left in place and was operated mainly as a way of providing a back up set of counters! The equipment retained the odds display dials but the system of odds clocks that must have been placed round the track had sadly disappeared. Apart from this the mechanism was in full working order and was well maintained up to the time of the last race. It disappeared in a week - though our Science Museum has preserved a small and representative sample of each piece of equipment at its Wroughton stores near Bath.
Our recording was restricted to noting the working arrangements of the machinery, taking photographs both b&w and colour slides and making a short video on its operation - it goes through two cycles.
I was interested in the machine primarily as a form of computer - certain claims can be made for the Julius machinery - such as its ‘time-sharing’ ability though, of course, Julius thinks in very different terms. I was not able to trace any influence from Julius machines to modern computer developments, sadly. I had hoped there might have been some direct inputs. Reading one of Julius’ papers I was intrigued to find that he had discovered Babbage - but here again it seems to be after Julius had done the bulk of his work, and he does not think very highly of him.
Not surprising one could argue - Babbage rested in the luxury of Government funding to build a machine that no-one quite knew what use it would be - by contrast Julius was building a highly commercial machine and needing to develop and expand its utility to meet the ever growing demands placed upon it.
Some of these thoughts have been presented to GLIAS and the Computer Conservation Society here in London. The GLIAS lecture was re-written as an article - which I would be very willing to send you. Snail-mail would be best as the source was never fully electronic and the delicate patent drawings from Julius’ UK patents would not survive scanning very well.
It would be very helpful if you knew of any sources of footage of film or video of Julius machines - we organise a film evening together with the Newcomen Society and the subject this year is likely to be early computing machines.
A definition of the word Computer from Doron Swade
On 2 October 2015, I was very fortunate to receive a definitive description of what the word computer means. I received it from an eminent expert on the history of computers and computing Dr Doron Swade MBE. Doron has had a long association with the Julius Totalisators and wrote a very informative article on them titled A sure bet for understanding computers, in New Scientist magazine, 29 October 1987, P49. I think his interest in the Julius Tote probably started when he worked for the London Science Museum where he became the Assistant Director and Head of Collections. He provided me with this definition when I requested one for a speech I was writing for the Institution of Engineers award ceremony relating to the award of an International Marker for the Julius Totalisators being held at Eagle Farm Racecourse. Following is Doron's definition:
What is or is not taken to be a ‘computer’ has changed over time. It is well known that the people who did low-level repetitive calculations in the 19th century and earlier right up to the 1960s were called ‘computers’. The features of the system described by John Von Neumann in his 1945 paper served as a ‘definition’ of a computer for a goodly while and to some extent still does – internal stored program, serial fetch-execute cycle, separate processor and memory, input and output. Turing’s famous paper from the late ‘30s has provided a more formal definition i.e. a thing is a computer if it is a Turing Machine – this is a logical definition in terms of a theoretical machine, rather than in terms of internal architecture and system features as in Von Neumann.
So in terms of language usage one can say that what it is the term ‘computer’ refers to depends on the historical period and the usage at the time.
Especially since Turing the medium of implementation (mechanical, electrical, electronic . . .) would not be considered a defining feature of a computer i.e. the medium the logic is implemented in would not be considered a material to a logical definition. Taken to extremes, one AI commentator referred to the ‘beer-can theory of consciousness’ meaning that the logic was all that mattered and it was irrelevant how this was implemented. So if the logic was implemented using beer-cans and computers were mind-like then we arrive at the absurd situation of needing to predicate consciousness of a collection of beer-cans. Having said which, in current common usage, a computer is now pretty well taken for granted to be electronic.
So we would not now call the Totalisators computers when the term is now understood to convey ‘programmable’, ‘electronic’ and ‘Von Neumann architecture’. The quote you cite from the New Scientist article judiciously uses ‘computation’ rather than ‘computing’ and I have used this distinction to differentiate between computing as manipulating symbols according to rules on the one hand, and doing numerical calculations on the other. The Julius machines, for all their remarkable features, could not be regarded as programmable symbol manipulators. They essentially do numerical calculations (mainly cumulative addition and division) and I have used ‘computation’ rather than ‘computing’ to make this distinction.
|Previous page||Go to the index||Top of the page|