MODE 3 with PMASK

Hi,

I'm trying to run a refinement in MODE 3 using PMASK="0 0 1 0 0" and DANG=90.0 in Frealign v9.11.
While the vast majority of particles behave as I expect (changes in the angle PHI are multiples of 90 degrees, with some deviations due to the subsequent refinement in MODE 1), a few particles end up having changes in the other parameters, which should be blocked by PMASK.
Interestingly, in these cases the blocked parameters seem to "acquire" the values from the particle located immediately before it in the .par file, which makes it even more strange.

Here is one example:

particles_9_r1.par - initial .par
C PSI THETA PHI SHX SHY MAG FILM DF1 DF2 ANGAST OCC LogP SIGMA SCORE CHANGE
25 319.30 351.66 150.65 0.25 -0.31 37313 1 7462.5 7979.2 -28.31 100.00 -4171 0.3784 11.41 0.14
26 78.94 47.53 166.04 11.79 10.35 37313 2 8178.6 8702.5 -26.14 100.00 -3954 0.3746 4.30 0.06

particles_10_r1.par - after running 1 cycle with MODE 3 using PMASK="0 0 1 0 0" and DANG=90.0
C PSI THETA PHI SHX SHY MAG FILM DF1 DF2 ANGAST OCC LogP SIGMA SCORE CHANGE
25 319.30 351.66 330.79 0.25 -0.31 37313 1 7462.5 7979.2 -28.31 100.00 -2433 0.2998 12.90 1.49 - OK!
26 319.30 351.66 331.46 11.79 10.35 37313 2 8178.6 8702.5 -26.14 100.00 -2344 0.3007 5.07 0.77 - ???

From cycle 9 to cycle 10, particle #25 behaved normally, but particle #26 had its parameters PSI and THETA changed besides PHI, and they took the same values as particle #25.

Any idea what is going on here?

Thanks!