Eph 6.21 in the NA28 has probably one of the most complex things ever seen in an apparatus. Whereas NA27 simply cited Sinaiticus in support of the txt at the word order variant: PANTA GNWRISEI UMIN TUXIKOS; NA28 now has: ALEPH in support of the same reading, but qualified with: ‘(*.2a).1.2b’. This, while entirely accurate, is possibly over-kill - offering four different slices of the history of the manuscript all ostensibly in support of the same variant; but the situation is interesting.
It is pretty clear that the original of Sinaiticus [labelled * in NA28] had PANTA U (which was caught as a mistake in the act of writing and dotted or crossed through [labelled 1 in NA28; S1 in SinProj]) then GNWRISEI UMIN. A subsequent corrector [labelled 2a in NA28; Ca in SinProj] added MIN between the lines (correcting the text towards the Maj. text, but leaving the following UMIN in place, resulting in PANTA UMIN GNWRISEI UMIN TUXIKOS). A further corrector [labelled 2b in NA28; CA in SinProj] rubbed out the MIN, and either dotted or crossed through the U.
It seems to me likely that the exemplar of Sinaiticus had the word order GNWRISEI UMIN, if it was otherwise there would have been no reason to stop and self-correct. Thus Sinaiticus shows the type of word order variation originating independently. So Sin* is rightly cited in support of the NA28 reading. Possibly Sin2a really meant to correct the text towards the Maj. reading, and could possibly have been cited on the other side (but within brackets). But probably the simplest solution would have been to leave it as it was in NA27! It is certainly one that will now take a bit of explaining to students.