Hi Mike. Still out of town. Yes, that is probably true. MRC files are not designed to store stacks of images. There is only one header for the entire file. The idea to store many individual images packed into a 3-D volume was a hack I think David Agard first initiated many many years ago. There are many problems still with this approach, despite the fact that it's become widespread. If you have only one header, should that represent information from the first image? the last image? the entire stack treated as a volume? There is no single answer to this question because there still is no "official" way to store stacks like this. For this reason, EMAN doesn't make use of most of the header parameters itself (they can't be trusted), unless they are things like A/pix, which are the same for all images in the stack (presumably). It does try to preserve the contents of the header for other program's use, and to generate reasonable values, but in the absence of a formal specification... There is also the issue that, due to memory concerns, EMAN works hard to NOT keep entire 3-D stacks in memory at once (these may be 10-20 GB in some cases), meaning the statistics on the entire stack aren't readily computed unless you read through all of the data in the file (very very time consuming).
Anyway, yes EMAN probably populates the header with values from the last single image written to the file at the moment. There is a group of developers (myself included) trying to formalize the MRC specification for stacks, but it is not yet done.