Juggling the phase of the E-field and subtracting from the H-field
to max the long path signal, then doing some cross-correlation
with the much stronger short path signal, then subtracting an
auto-correlation of the short path signal, gives

 http://abelian.org/vlf/tmp/nwc140423a.png

The peak at a delay of about 42.2 mS agrees quite well with the
path difference of 12708.9 km ( 26369.52 km - 13660.62 km).

12708.9 / 300e3 = 42.36 mS

The MSK signal isn't ideal for this sort of thing, given its
strong and broad auto-correlation sidebands.

--
Paul Nicholson
--