What is the purpose of the spring used in point-contact transistor?

Here are some experimental results I have obtained with a galena detector.

Preamble

Galena detectors were the predecessor of diodes. It was composed of a galene cristal and a needle of copper in contact with it. So far I know galena detectors always have a compression spring.

Here is such a detector similar to the one I have tested. One can see the spring.

The tests were done with a curve tracer.

Results

To obtain a detector effect, the pressure must not be too much nor too light.

Thanks to the spring the operator is able to master this pressure. For this reason, the spring is clearly essential

The best results (ie non-linearity) are obtained when one first apply a little bit to much pressure then release progressively the pressure until a good non-linear curve appears on the screen of the curve tracer.

I don't have the theory behing this. Maybe the first pressure flattens or cleans the surface (oxyded ?), but I seem to recall that when one apply once again more pressure after having obtained a good result, things are worse.

So the modelisation of this kind of problem seems to be complicated. There are some effects with memory that can improve the results.

You need enough force to maintain contact, but not so much that you crack the Ge crystal.

If you just directly connected the contact to the structure shown holding the spring, with no flexibility, it would be very easily to apply excess pressure and damage the crystal.