Proc distance method=dgower

Can anyone please explain what SAS is doing with the missing data when computing Gower or dGower coefficients? I have gone back through my data and computed some of the coefficients by hand (as outlined in Robert Drennan's 2009 Statistics for Archaeologists) but I am not getting the same numbers given in the output matrix. Drennan gives a score of 1 for any matching nominal variables, 1 for anomimal positive matches, and for ordinal takes the absolute value of the difference of the values divided by the number of ranks for that variable; all variables with missing data are excluded case by case. SAS is doing something different. I think it has to do with the standardization of variables or its treatment of missing data.
My syntax is as follows:

title 'Similarity of Cases';
proc distance data=work.thesis out=work.dist method=dGower;
var nominal (lamcon--torsom3r)
anominal (stron--conabm3r)
ordinal (lamext--radm3r);
id Caseid;

Any ideas?