1.8.2

One fix, and it turned out to be larger than it looked.

The thermometer could not give a figure to anybody, ever

You play it perfectly and the line under the reading says "held long enough and it would have given a figure". It was held long enough. Nobody has ever got a figure out of it.

Three reasons, and the first is arithmetic.

It was reading the wrong number. The question "was it shaken down?" was asked at the end and answered from the height of the column - and the column is reused during the hold to show the new reading climbing. So by the time anybody asked, the thing being divided was the new reading rather than the old leftover, and the test ran backwards: the longer it was held - the better the reading, the more careful the player - the higher that column stood and the more certain the game became that the thermometer had never been shaken.

Run out over every hold and every starting leftover, the old code could return at most 0.300, and the server wanted exactly 1.0. The fact is now remembered at the moment it happens.

The quality was being multiplied by the steadiness of the hands. The room, the instrument, the assistant, fatigue, sleeplessness - all of that multiplies into the quality of any treatment. But a reading is not a treatment: a tidy room and a rested pair of hands do not move a column of mercury. A flawless reading taken in a field still could not reach one. The thermometer now gets what the screen earned and nothing else.

And it was compared against one exactly. A float that has been through a packet and a clamp is a poor thing to ask equality of, so the threshold now sits just under.

The second line stopped lying while it was at it: if the column really was somebody else's, that is what it says now, rather than telling you to hold it longer.

A figure comes out after 4.7 seconds in the fold.