CALC-089

SRC-085 Figure 17's own column total arbitrates two defects in its own table

Supports CLM-0871. Ten two-digit multiplications and one column sum, given in full so anyone can redo them in a minute. The point of this calculation is that it needs nothing from outside the figure.

Read from a RENDERED image of PDF page 44, printed page 32. The text layer of this scan returns the figure as garbled fragments and does not carry the Klystrons/Subarray column at all, so the table below could not have been obtained by extraction.

The arbitration

Step 3. 644 x 24 = 15,456 and the cell prints 15,458. Substituting 15,456 makes the column sum 101,552 EXACTLY, which is the total the figure itself prints. So the printed total agrees with the product and disagrees with the printed product cell: the PRODUCT CELL is the defect.

Step 5. 784 x 18 = 14,112 and the cell prints 12,544, which is 784 x 16. Substituting 14,112 gives 103,122 against the printed column, or 103,120 if the step-3 correction is applied as well, and the printed total is 101,552 either way, so the total agrees with 12,544 and therefore with a multiplier of 16: the MULTIPLIER CELL is the defect, not the product.

Different cells in the same table, and the figure's own total settles both without any appeal to physics.

BOTH FIGURES ARE PRINTED ABOVE BECAUSE THE FIRST VERSION PRINTED ONLY 103,120 AND SAID IT CAME FROM SUBSTITUTING INTO THE PRINTED COLUMN, WHICH GIVES 103,122. The script had folded the step-3 correction in and the prose had not. Caught by the independent verifier of CLM-0871. It changes nothing - both miss 101,552 by more than 1,500 - but a number in prose that cannot be reproduced from the operation the prose describes is exactly the defect this file exists to expose, and it was sitting in the file arguing that a 1978 table's arithmetic does not reproduce.

THE ARBITRATION DOES NOT MERELY ASSUME THE TOTAL IS RIGHT, which is the objection it was checked against. The row-level inconsistencies at steps 3 and 5 are established by direct multiplication BEFORE the total is consulted, so the total adjudicates between hypotheses that are already live rather than creating them. And at step 3 the single-cell fix is essentially forced: 15,458 = 2 x 59 x 131, divisible by neither 644 nor 24, so no correction to the subarray count or to the multiplier can produce it. That argument is the verifier's and is stronger than the one the claim was written with. A third route agrees independently: the blind forward computation in CALC-088 reaches 6.749 GW radiated with 16 klystrons per subarray at step 5 against the document's stated 6.72 GW, and 6.869 GW with 18 - so the radiated power picks 16 as well.

Sabotage checks

The script that produced this asserts each step of the argument and would abort on any of them: that the printed column sums to 101,554 and NOT to the printed total; that correcting step 3 to 15,456 makes it sum to the printed total EXACTLY; and that correcting step 5 to 14,112 instead does NOT. Change any printed value and at least one assertion fails.

The step boundary radii, recorded because they exist nowhere else in this register

93.9, 177.3, 229.5, 271.2, 323.3, 365.1, 396.3, 417.2, 458.9 and 500.6 m, printed along the top of the array diagram on the same page. They are the quantisation of the 10 dB truncated Gaussian taper and are only legible on a render. Note the outermost is 500.6 m, so the array is marginally over the nominal 500 m radius of a 1 km aperture.

Cite this