Digital BW, The Print

I have used the spectrocolorimeter datacolor #1005 of my PrintFixPro
package for 2 years, profiling about 40 different papers for color and
grayscale with good results. But, just in the last days, measurements
of dark patches seem quite strange on the color coordinates in Lab
values. For example, here 5 Lab measurements of the step 96 in a 51
steps grayscale, waiting about 5-10 second between them:
1) 7,99 | -2,91 | 1,26
2) 8,25 | -1,29 | -0,9
3) 8,11 | -0,85 | -1,1
4) 8,29 | -0,22 | -2,13
5) 8,00 | -1,47 | 1,28

B value range from 1,28 to -2,13 and I can notice an evident loss of
contiguity, for a and b values for dark patches, comparing same paper
profiles created in the last days and one month ago.
I have repeated the same tests with different sw upgrades, from
PrintFix 1.0 to Spyder3Print, on different computers and different usb
ports, but nothing to do.
Have you any suggestions?