General Report on Tunny
24B Page 118
(e) Robinson rectangles.
Two tapes are used, viz.
(i) cipher tape, on bedstead A, tape length one less than a multiple of 1271, with start and stop.
(ii) control tape, on bedstead B, tape length 1271, with a start (for counting position only) and a single "E" in the first place of the tape.
Switch B = "E": this selects those places on A opposite the "E" on B, which are of course spaced at intervals 1271 (the length of B). Moreover in each successive revolution of the cipher tape A, all the places opposite "E" will move one forward (because the length of A is one less than a multiple of 1271), so that the 1271 cells are selected in "diagonal" order.
The score counter is split; one half counts ΔZ1 + ΔZ2 = . the other ΔZ1 + ΔZ2 = x.
The difference between them is the score for the corresponding cell.
Their sum is the depth, which serves as a check, for there can be only one change of depth.
The position counter is split to repeat after 41, 31. Since the start on A is used, it necessarily records how much A is ahead of B and so runs backwards, 0000, 4030, 3929, ... These figures are written along the sides of the rectangle to check the entering.
As a check on Robinson scores the machine is allowed to run round a few times after the rectangle is finished; it immediately begins to repeat the rectangle.
In fact, the B tape contains also 41 "4"s at intervals 31, which are used analogously for 
runs. The first "4" is one place back from the "E": this is merely a trick to make the position counter readings tally with those of a run on Colossus.
For the details of plugging and switching see Synopsis of Robinson plugging (54J). It will be noticed that some unnecessary cords are used: this is to minimise changes between 1 + 2 rectangles, , and counting "9"s.
(For early versions see R1, p 32.)
Back to General Report on Tunny. Contents.