1974 · Reprogrammed Scientific Calculator · Work 946 of 1,871
A kitchen table in a terraced house in England on a winter evening, a self-assembly kit just finished, an upright pocket calculator of cream-white plastic about eleven centimetres tall, at its top a dark violet-black window with a soft red glow, below it a recessed dark key field with pale blue keys, light grey keys and a pale blue slide key, every key plain smooth and unmarked, a hand in a brown jumper sleeve holding it at a slant above a tea towel, a soldering iron resting on a wire stand, a reel of solder, a saucer with three tiny screws, a mug of tea, warm light from a standard lamp.
What it was used for
Sinclair Radionics launched the Scientific pocket calculator in 1974; according to Ken Shirriff's analysis it sold for 119.95 dollars or 49.95 pounds, while the HP-35 went for 395 dollars. It computed trigonometric functions, logarithms and antilogarithms, working in scientific notation. The price gap was paid in time and digits: sin 0.1 came back in under 1 second, sin 1 took about 7.5 seconds, arccos 0.2 about 15 seconds. Its scientific operations, Shirriff writes, "only have three decimal places of accuracy at best", against 7 to 11 digits on the HP-35 designed to a one-second limit per computation.
What was new
What was new was not a function but a frugality. Shirriff read out the chip: a TMS0805, a part meant for a plain four-function calculator, whose read-only memory holds 320 instructions, by his account half of what sat in the HP-35. The scientific functions were written into those 320 words instead of being given circuits of their own. The code was written by Sinclair together with Nigel Searle, a programmer with a mathematics PhD, "in a few days in Texas"; Hewlett-Packard, the same source says, needed "two years, 20 engineers, and a million dollars". The barb: the saving came out of accuracy, in a place the buyer noticed only while calculating.
What the sources leave open: Technology, prices and timings come from Ken Shirriff's article on his reconstruction of the chip (the timings, he says, measured on a real calculator), not from a company source. The Science Museum Group confirms the 1974 date and Sinclair Radionics as maker but gives no price or technical detail. The later 9.95-pound kit Shirriff mentions has no second source and is left out here.