I tested the profile. It works, but on the Lab profile visualization (ColorSync, in OS X) it looks like it is floating above the intersection of the axes. [...]The profile created by Datacolor does not have this (nor the generic profile I have been for this paper). [...] is it normal to have profiles that look like that?
Yes. There are several reasons:* Colorsync Utility apparently uses the A2Bx tables to create the visualisations
* even worse: Colorsync Utility always uses the A2B0 (i.e. the perceptual intent) when you compare profiles. I.e. What you compare are not device gamuts, but at best gamut mapping spaces.
* some vendors use the inverted B2A0 table as A2B0 table, i.e. they try to "undo" the gamut mapping in the A2B0 table. In such profiles, the black point looks like L=0 in Colorsync Utility.
* In contrast, Graeme decided to make all A2Bx-Tables colorimetric (a very reasonable decision IMHO). For details see <file:///Applications/argyll/Argyll_current/doc/iccgamutmapping.html>. When you visualize the A2B0 table of an Argyll Profile with ColorSync Utility, you see the "real" (relative colorimetric) black point of the profile, which has always L > 0
* when you use the Argyll gamut visualization tools (iccgamut, viewgam), you can pick such things mor detailed to pieces than with Colorsync Utility. Admittedly this is much less convenient than double-clicking a profile in MacOS.