I understand PD4ML cannot merge documents and the recommendation is to use another product for merging or to string multiple HTML documents together then convert it with PD4ML. If I do the latter, is it possible to set the start page number for each logical document?

You may define that by PD4ML API call PD4PageMark.setInitialPageNumber(int)

See: http://www.pd4ml.com/api/org/zefer/pd4m ... eMark.html