Early 1994 it became more or less clear why we found a different result in our source counts than other authors. The result of the source counts depends a lot on how you define a source. If a source finding program is fitting a source with several components then it is very important to know how these components are combined (if they are combined at all). Since no author really described how he did this we think that nobody has realised how big the effect on the source counts can be. We have done some tests with data taken by Wieringa and were able to reproduce both Wieringa's result and our result for the source counts depending on the way we combine source components.
In June 1994 Rengelink discovered a flux problem with sources below 100 mJy.
This problem is still under investigation. Suggestions by de Bruyn that
weak sources are not taken in the selfcal solution and therefore have a
sinc function response (due to the FFT of zero points in the plane)
instead of a gaussian response (from the restoring beam) are ruled out by
Rengelink. Fitting a sinc function with a gaussian estimates the flux in the
wrong direction, it would only makes this problem even worse.