Mark, thanks for providing error margins. I think that this means that the accuracy when measuring delta-T, is ±0.1°C, but it's accuracy without calibration is ±0.4°C. Therefore thermometer was calibrated that boiling point was 99.7°C. Henceforth thermometer measures accurately.
To add one important detail to my previous post, that Mats collected water into blue bucket and did not let it flow to the sink like when Steven was observing Rossi's test. I think that most of the water that was overflown, was before reactor kicked in and started boiling the water. —Jouni