Еще от Дюка:
-------
This method therefore both assumes and requires a good linear fit, but it is clear from Figure 1 that the data are not at all linear (the linear correlation is about –0.003, even with α Cen, θ Cen, and β CVn omitted) and the fit explains essentially none of the variance in the data. This is particularly critical in this case, because the method of Dambis and Efremov to estimate the uncertainty in the slope b, and hence the epoch T, effectively assumes a good quality of fit.
......
If a linear fit is still required (!), though, then an alternative, and greatly preferable, procedure is to compute the estimated sample variances in ∆2β and ∆2λ cos β for all the stars in the catalogue, excluding outliers, and to use these as the weights in the least squares fits. This does not overcome (!!!) the objection that the data are not linearly correlated (!!!), but it does remove the need to use the scatter in 40 or 80 points to estimate the uncertainty in the slope.
----------------------
Далее Дюк косвенно "подтверждает", что Альмагест написан не в 10-ом веке:
Tβ = -188 ± 243
Tλ = -82 ± 317
Tλβ = -142 ± 195
т.е. оценки куда грубее у Дюка (пытавшегося исправить Ефремова с Дамбисом), но, мол, не настолько что бы "потерять" опровержение Фоменко. Только Ефремов забывает о неправильности линейной модели, указанной Дюком вверху.
П.С. плюс еще спорное отождествление о2 эридана... но об этом писалось много.