Dear Dr. Baxter and other dog-interested Zooarch's. 

I also used Harcourt's factors to determine dogs shoulder heights from Chukchi Peninsula (archaeological and modern material). I had some another picture with incongruity of heights calculated by different elements. I attach scatterplot illustrating the calculated heights by different elements of some complete skeletons of modern and archaeological dogs from Chukotka. So it seems to be the other type of problem with H. factors. The Wilcoxon Matched Pairs Test showed statistically significant difference between heights calculated by proximal (humerus and femur) and distal elements (tibia and mean value between radius and ulna): (N=19, T=5,5, Z=3,6, p=0.000316)

I won’t wonder if somebody else shows any another kind of this picture J

I suppose that very great differences between skeleton proportions of different dogs (ex. Harcourt used and my Chukchi dogs) may be the main reason of such different discrepancy of calculated heights by skeleton elements... So maybe we should tune Harcourt's factors for either "dog types" if it's possible?