Dr. Elton James Dyce documented this and says that some of the water comes out when some of it crystallizes. The risk is that the liquid part may end up with a high enough water content to ferment. It may not, if the water was low enough to start with.

http://krishikosh.egranth.ac.in/bitstream/1/2031942/1/19310.pdf
https://www.tandfonline.com/doi/pdf/10.1080/0005772X.1958.11095058

 

The saturation of sugar in water is temperature dependent, why "heating honey reliquifies it" . As honey crystalizes, the sugar drops out, but the water remains in solution, the solution, the "mother wort" will equilibrate with the storage temperature right at saturation for that ambient temperatue. The prescence of "seed crystals" in the jar will make crystalization more likely (as the seed crystals lower the sugar concentration required to form and grow additional crystals).

A honey that is crystallizing and stored cool, as in a garage in the wintertime, will on warming to spring ambient temperatures fall below critical fermentation-safe concentration and ferment from wild yeast. The dilute "wort" will not form additional crystals, as the solution is below saturation for the ambient temperatures.

 

Interesting! I was starting to wonder if there was somehow two distinct nectar sources that would not blend, but this makes more sense. The liquid part does seem pretty thin.

 

This is the worst thing about raw honey, and IMO, a major reason why until recently most honey sold in the USA was 'pure honey'. Heating or worse, pasteuriziing, the honey stops this or at least delays it. This makes for a more attractive product to sell. Appearance is everything.
It happens to some of my raw honey when the jar contains a mixture of 2 very different honeys. It turns off the customers.