1) All the lines of the NI and the 3939 Hz line of the WI have a temperature dependence of around 0.3 Hz/K, the 5642 Hz line has 0.6 Hz/K and the 5652 and 5660 Hz lines move with only 0.1 Hz/K. This might indicate some fundamental difference between the various resonances. Note that the data was taken during a thermal transient, more data is needed for accurate numbers.
2) The 3939 Hz line of the WI is a much cleaner signal than all the other lines, maybe because this resonance has a very high Q. (Note that the visible 'steps' are caused by digitization noise of the thermometer.)
