it is probably designed to be somewhat loose in the holder by design, so that it can expand as it absorbs the ehat it is suppose to do.
I am not familiar with the head in question. Is there any chance some part of the flange that surronds the filter has been bebt and is thus not quite as tight as it may have once been?
The other possibility - is there a bevel on the edge of the filter, and hence an up or down side?
My only experiencce with said glass in a loose state was in a beselar 23 many years ago.
I printed EP/2 with acetate filters, and the heat resistant piece sat in the top slot of the filter drawer, and the acetate stack lived in the bottom stack.