I performed data collection of a single crystal at different temperatures. The measurement was inserted as a temperature sequence with an automatic change of measurement temperature in CrysAlisPRO. The measurements were performed in the temperature range of 300-410K. The strangest thing was that after data reduction in measurements performed at 360K and 390K, the completeness was lower than 100% (82 and 88%, respectively). This is what the statistics look like in CrysAlis PRO for data collection at 390K

resolution kept theory unique complete redundancy F2 F2/sig(F2) Rint RsigmaB

inf-1.87 563 281 272 96.8 2.1 5525813.24 48.73 0.016 0.012

1.87-1.45 471 290 272 93.8 1.7 1088712.53 19.94 0.019 0.027

1.45-1.26 706 295 272 92.2 2.6 324223.86 16.41 0.053 0.048

1.26-1.13 693 310 272 87.7 2.5 422142.86 17.53 0.040 0.038

1.13-1.05 684 308 272 88.3 2.5 330071.30 14.62 0.045 0.044

1.05-0.98 620 293 272 92.8 2.3 186867.17 10.55 0.057 0.059

0.98-0.93 582 305 272 89.2 2.1 96963.70 7.34 0.071 0.080

0.93-0.89 551 302 272 90.1 2.0 59217.36 5.66 0.094 0.101

0.89-0.85 521 306 272 88.9 1.9 39703.64 4.51 0.122 0.123

0.85-0.82 443 388 273 70.4 1.6 18189.50 3.05 0.174 0.212

inf-0.82 5834 3078 2721 88.4 2.1 789286.10 15.12 0.024 0.022

What could be the reason for this? After data reduction performed for data obtained at the remaining temperatures, the completeness was over 99.5%.