NavList:

Frank

Your idea of using the sextant has been well discussed and does have merit.  The ideal sextant can be set to any arbitrary angle.  All you need is an ideal sextant or one in which all errors are *known*.  Then all of the observed errors can be attributed to the measured sextant. Failing that, the truth remains hidden.  The observed errors will be a hodge podge of errors in the ideal sextant and errors in the measured sextant.

The beauty of an ultradex is the precision it affords while permitting easy manipulation.  Here's one on eBay right now http://m.ebay.com/itm/191042953903?nav=SEARCH&sbk=1 the top plate and bottom plate engage with a patented interface, that guarantees accuracy and repeatability.  It is an independent NIST traceable, calibrated instrument.  That is something simply not achievable with your sextant concept, attractive as it may be.

The top and bottom plates of the ultradex  are set parallel to the plane of the sextant.  The beam splitter (or prism or mirror) mounts perpendicularly to that, just like an index mirror.  So rotating the ultradex top plate rotates the horizon optical path. 

Take a look at this arrangement
http://www.globalspec.com/reference/14523/160210/chapter-2-2-beam-splitter.

In place of the collimator put the sextant to be measured.  In place of the mirror 1, is the horizon path disturbed. In place of mirror 2 is the horizon path, undisturbed.

Perhaps this will engender a more meaningful response

Brad