> On the other hand, these entries in the Lexicon Poeticum
> Antiquæ Linguæseptentrionalis [1] would suggest that there
> must have been some textual support for <skýja>, unless
> subsequent scholarship has discredited or corrected these
> readings

I’d guess that that’s what’s happened, given the complete
lack of any mention by GPH, but it would be nice to see all
of the mss.

> I must admit that, for the only ms I have looked at, it
> looks more like <skytia> than  <skyja>
> (https://handrit.is/en/manuscript/imaging/is/AM02-0155#page/10v++(28+of+79)/mode/2up

Nice legible hand, that! And I don’t think that there’s any
doubt that it’s <skytia> here, though the scribe does seem
to have divided it improperly from the preceding word. I
make it:

Margr mundi sa sender sokn gifrs muni lifi liNa vangs f^
longu lundur er nu rekr undañ ef brynh^þ^ bædi biti
tygiligs kytia nidr drepur skalld und skilldi skeGi mal
sem eGiar.

I don’t know the hand well enough to expand the three
superscripted scribal abbreviations, which I’ve simply
replaced with carets. I am familiar with the use of upper
case <N> and <G> to indicate <nn> and <gg>, respectively.
The sequence <nu> is in several places on the page
indistinguishable from <nn>, save that the latter seems not
to occur, being replaced either by <N> or by <ñ> (where the
tilde actually represents an overbar). (By the way,
<undann> is a known spelling variant of <undan>.) Vowel
length, not too surprisingly, isn’t indicated, and neither
are <d> and <ð> (which is completely unsurprising).
Finally, the scribe attached the <s> of <skytia> to the
preceding word and left a bit of a space between it and the
<k>, but clearly <tygilig skytia> is intended.