On 12 Feb 2010, at 18:00, cctalk-request at classiccmp.org wrote: Thanks to all who responded, I think you will like it even more when you see the new layout with 8 decks in a row, four of which worked at last years open day. The sound is produced by executing a very small loop of three or four instructions many times which contains a (decimal) multiply instruction, the duration of which varies with one of its operands. Zero takes almost no time, 555555555555 takes longest because for each digit it can either count down for 1,2,3,4 or 5 or count up for 6,7,8 and 9. There is an increment in the loop hence the time varies with each iteration. A conditional branch at the end of the loop flips a bit which drives the speaker. In the demo software being run in the video there is also a small test for the bottom 2 or 3 digits of incremented value being zero which allows it to branch off and drive the peripherals from time to time. This does no seem to affect the sound too much though it is not quite as pure as the original 'ghost' program. From time to time (but not in the video) it does a bubble sort of a block of data read from tape which produces a more recognisable 'computer' type sound. I once keyed in a short sequence of jumps activated by some of the switches on the front panel. The longer loops worked fine but there seemed to be a problem with the top note. There happened to be a young lady present and she said she could hear it fine, it seems it was simply beyond the frequency range of my old ears. I think it would flip the bit 1000000 / 12 times a second, so dividing by another 2 that would be 41.666 kHz. The machine has a nominal 1MHz clock derived from a 250kHz timing track recorded on the last drum accessed and the shortest instruction is 12 clock cycles. As another thread is discussing cats, I have to add that the first time I saw my 1301, there was a cat asleep on the main power stabiliser rack which was stuffed full of 6 inch by 4 inch by 4 inch heat sinks for the GET875 transistors used in parallel to regulate the various voltage DC supplies. It was the rack which was stuffed, not the cat by the way, the cat got up later and played with my host's hands as he tried to operate the main control panel. You've got it, though I didn't know they ever shipped them with just 400 words. I knew it was theoretically possible to have just one 'barn door' but that must have been incredibly restrictive, especially on a tape machine where you need I/O buffers for Data Transfer Unit to do what we now call DMA into/from. Yes it could be programmed in MPL (Mnemonic or machine code or TAS (Thirteen hundred Assembly System) or MAC (Manchester Auto Code) and later there was COBOL and ICT's attempt to make COBOL more bearable, RapidWrite. Do you mind telling me where this 1300 or 1301 was? I don't suppose you know its serial number do you? I am trying to identify as many machines as I can, especially so I can find out how many were made. Mine is number 6 and I have parts of numbers 58, 75 and 166 but I suspect the number nearly reached 200 but the best remaining official ICT record is incomplete, some pages were lost and it was a marketing document, a list of customers in alphabetical order, and probably made before production ceased anyway as marketing would hardly be interested in an obsolete product. Your 400 words reflects on another thread where someone said "In the mid 60's only large companies had systems with greater than 16K memory and disc drives. Mag tape 800 and 1600 bpi if you were lucky was the norm." Taking the middle of the mid 60's, 1965, about a quarter of the machines in the UK were 1300 series with between 400 and 2000 words of Immediate Access Store (core), no discs just usually one 12000 word drum. Those lucky enough to have mag tape (which roughly doubled the cost of the computer as well as requiring air conditioning) were 300 bpi, usually 10 track (4 data + 6 CRC) half inch or for a lucky few, 16 track (8 + 8) one inch wide tape. There were of course scientific machines like ATLAS around but only three? were ever built, most actual data processing was done on much more mundane machines like the 1301. The IBM 360 was announced in 1965 but how many actually got their hands on one in the UK that year? If you only had 400 words (4800 digits) of storage, who would waste it hold 19s. As to using PICTURE XX , I only ever wrote one COBOL program (put me off for life) but I think PICTURE XX means two characters, which means 4 digits so exactly the same storage size as PICTURE 9999. Roger Holmes.