I fully agree that the Gulf of Merofro is proportionally small compared to Hudson Bay, but the details inside closely match those of Hudson Bay. For instance, it shows Southampton Island with the Bell Penisula, Coats Island and Mansel Island, along with other coastal inlets. Points F, G and H on my Figure 1.51 are easily identifiable, but are clearly in the in the wrong location on Mercator's map. I do not know which sources Mercator used to compile his maps - possibly verbal descriptions only.
On the issue of conjecture, any form of 'educated guess' requires at least some a priori knowledge, otherwise there should be no correlation between what is conjectured and reality. This principle is what convinces me that the continent had been mapped before the 16th century. Mercator's map of Africa shows lakes in the Sahara desert, the existence of which is supported by NASA maps (see Figures 31-39 here (
http://www.riaanbooysen.com/terra-aus/35-proof?start=11). These lakes could only have existed probably 8000-10000 years ago. How could Mercator have guessed the existence of these lakes at all, and then reasonably accurately as well?
Terra Australis likewise is believed to be nothing but conjecture, a land mass believed to exist because it had to balance the land masses in the north. This is all very well, but two other forms of the southern continent also exist (see below), which in my opinion eliminates the possibility of pure conjecture.

How could anyone have imagined these to exist shooting in the dark?