I would venture that generally speaking, two orchids for which 2n "consistently measured in nature" equals different numbers then they are not part of of the same genus. Doritis pulcherrima and Doritis buyssioniana belong to the same genus because though the first consistently measures 38 chromosomes and the second consistently measures 72, the first occurs in nature at 2n, while the second is almost always 4n.

The problem is, I believe, different specimens of the same species are measured many times at different chromosome counts, independent of poliploidy. This makes it harder to classify.

In any case, this is not a problem in the Cattleya grouping(2n=40) or the Vanda-Phal grouping (2n=38). There you have tremendous consistency. Taxonomists keep creating new genera in Cattleya (Guarianthe, Cattleyella) and Vanda (Amesiella) but based on things other than chromosome count. Within these groupings orchids may not breed together (i.e., phalaenopsis with paraphalaenopsis) perhaps because of the different "sizes" or "make up" of each chromosome, rather than the number of them. W. W. Moir observed difficulty in crossing Cattleya types of different altitudes and speculated the sizes of the temperature-endurance chromosome might vary, making a match difficult.

So in other words, for breeding purposes Cattleyas and Vandas are great as long as there is no poliploidy (which is frequent in awarded hybrids). Most of the others groupings are "weirder," if I may say so.

Lastly, it seems species with different chromosome counts can cross very rarely. This is bizarre. In 1976 Rumrill, who is known for exotic crosses, registered Notylidium Rumrill, a cross between Onc. sphacelatum pod parent [2N=56] and Notylia barkeri [2N=42]. The resulting hybrid appeared to have characteristics of either parent. Go figure.

I hope I didn't confure the issue more. :-)