SUMMARY
Yesterday's odd inclinometry was due to a switch set incorrectly.

It was speculated that a switch on the TMU inclinometer had been knocked during a recent TMU fix, and, indeed, the FILTER switch was ON rather than its usual OFF. This was set back to OFF and inclinometry performed :

   Datasets     HST start      mean leg temperatures   Humidity
                               start   middle   end       %
   inc990228      13:42         2.2     3.4     4.4      25
   inc990307      10:30         0.6     1.6     2.7      40
   inc990308      09:55         1.3     2.1     2.6      25

The difference between the model derived from the new data and the model of 990228 is shown below :

There is a sign of the spikes, which according to Justin's working model is due to a temperature differential, but generally there is fair agreement, with rms scatters in the (F1,F2,F3) differences of (0.72", 0.32", 0.36"). The concerns from yesterday are quelled.

Despite the fingers, the symmetry remains good (see the plot ), and the slope of the TY-vs-RY relationship of 0.82 is back to normal.

I have not installed the new track model - we'll be doing this all again on Friday.

For future reference : FILTER OFF, GAIN LOW.
I shall label the inclinometers with this notation asap.