inc970316_1 CW start 20:00 leg temps 0.8 C
                 mid 21:25           0.5
                 end 22:50           0.3

inc970216_2 CCW      22:55           0.3
                 end 02:00           0.0

Comparison of CW data of 970213 and 961027 (both 3C) show similar differences, but involving wheels 2 & 4 with this same set of joints.
Models of 970213 and 970228 (1C) were more similar in this regard, although 2" structures are seen. For comparison sake, the models of 961027 and 960826 show only noise.

Additionally, the CW-CCW of 970316 show huge, 10", spikes at these same azimuths. While only that at az=147 persists to the difference between CW and CCW models , there are other 2"-3" systematics present.

An explanation is sought for why these joints respond to some wheels and not to others, and why this should change with time :
If a wheel with heavy load induces a response from a joint, and a wheel without load crosses a joint without response, then following the central bearing adjustment on 970213, it would seem that wheels 2 & 4 were carrying more load than 1 & 3.

Q ? - is this possible ? The 970316 (CW) model , is installed.

17:50 inc970318cw  standard config  start time   17:50  19:15   20:40 end     
                                   mean leg temp -1.7   -2.0    -2.4

      Ch    1    2    4    5    7    8
           TX   TY   LX   LY   RX   RY
          0.2  0.3  0.5  0.4  0.4  0.2 "    (NB 970316 temperature +0.5K)
      So I have confidence in the accuracy of these CW runs.
      The 970318 model is NOT installed, since the prevalent temperatures
      are considered unusually low for normal observing conditions.

20:45 inc970318ccw  (-2.5K) is a part CCW run from az 160 to 130.

21:15 change Right A-frame inclinometer box to receive data from the 
      inclinometer on the radial arm between the central bearing and wheel 2 
      at the rear right. Adjust to bring typical X & Y readings to near zero.
      NB : the X-axis points radially outward,  Y-axis points CCW.

21:25 inc970318cwrad.dat started  (-2.5K)  ends 00:15 (-3.3K)

again gives these statistics :
      Ch    1    2    4    5    7    8
          0.2  0.3  0.5  0.4    -    -  " 
      but with  0.5" to 1"  excursions at several azimuths :
                        approx 95   120   140   275   300   325 
      corresponding to wheel    1    3     2     3     1     1  
      passing joints          9/10  3/4  14/1  9/10   3/4   4/5
   
00:20 inc970318ccwrad.dat from az 190 to 100 only (-3.4K). The difference
      between the CW & CCW radial arm data in this part circle 

 shows 1" to 1.5" differences at joints 3/4 and 4/5, but <0.5"
      effects at 5/6, (again, 20mV = 1"), while the data for the other
      channels confirm the 970316 CW-CCW data. Having this inclinometer
      on the radial arm to wheel 2 is perhaps not ideal for illuminating
      the large effects seen above that seem to involve wheel 3.
00:50 end

1. The CW data of 970318 confirm the CW data of 970316.
2. Large (5") changes at several azimuths have occurred at different times since 961026, some before and some after 970213. These locations involve all wheels (except #4 it seems), but favour joints 14/1, 3/4, 4/5, 5/6 and 9/10.
3. The 0.5" excursions between 2 contemporaneous CW datasets (inc970318cw and inc970318cwrad) reveal potential for variations on short timescales.
4. The absence of excursions between 970318cw & 970316cw may be fortuitous.
5. The differences between CW and CCW data are more extreme than those of 961026, with large and sometimes variable differences at several azimuths - the latter being at azimuth 147 where wheel 3 passes joint 4/5. The CW model is then less applicable to pointing in the north, although, as it happens, the moduli of the CW-CCW difference of model parameters (F1,F2,F3) near az = 360 are certainly <3".
6. CW-CCW differences are evident in the radial arm data, but the location of the inclinometer on wheel 2 (the rear right wheel) is not optimal for illuminating the more interesting current effects associated with other wheel/joint meetings.
7. model1-970316 (based on the CW data) has been installed. It is similar to model1-970318 but the latter was taken at temperatures unlikely to be encountered in normal observing.

Implications for Future work