I know this is a really old post, but Chris you do not happen to have a recording of this, do you? I was just going through recordings, sorting and filing them, when I came across what is probably the signal you are describing on a different date. 6923 kHz CW, dashes are 0.7 sec long, and leading edge to leading edge is 12.6 seconds, on November 11, 2013, at 0150 UTC.
The pulse spacing and length are potentially interesting. The dash length goes evenly into the spacing 18 times. This could be the multi freq Pips signal, or something related to it. That network has used 0.7 sec pulse lengths before, and it has used "unusual" freqs (outside its normal set) for some odd dash / spacing combinations.
T!