“…Putting measured (a-lx) into the multiplier (1-exp(1/2/(a-lx)) yielded the 720 standard deviation and hence uncertainty ef23 = {0.0%, 0.1%, 0.9%} depending on 1/2, the duration 721 of half cycle. Otherwise Equations [10] and [11] hold strongly coupled terms and the procedure was 722 not amenable to straightforward manipulation of variances, and instead the same simulation was 723 run 250 times with a Gaussian distribution of random errors (lines 24,25,26) …”