The graphs shown here are mainly for the benefit of the other HERWIG authors. They can be expected to change before the release of the next official version of HERWIG, so should not be relied upon by anyone else.
In each figure, the solid line is version 5.8c, the dashed line is Giovanni's improved version (currently called 5.8c+ within ZEUS), and the dotted line is my version (which I am calling 5.8d, but it is not the final, public 5.8d yet). The HERWIG Et flows use exactly the same cuts as the H1 paper, DESY 95-108. (You probably already realized that the numbers given as the x ranges should be multiplied by 10^-3).
There is also a comparison with the older H1 paper, DESY 94-033, which includes other distributions like the `seagull' plot.
Jiri Chyla also made some plots comparing my version with various values of the PSPLT parameter with eachother and with H1 data. He has also made the equivalent plots with v5.8 replaced by Giovanni's modified version.
The two modified versions are fairly similar to eachother, with mine producing somewhat more Et. It is clear that overall the Et flows are always too peaked compared to the data, not plateau-like enough. My version fits the newer data better, Giovanni's the older - it is not really clear to me where the difference between the two data sets comes from. In the only place in which they fail badly, the seagull plot, they are almost identical to eachother. According to Jiri, this cannot be improved by tuning (I have left all parameters at default values).
The overall conclusion? I still vote for my solution to the problem - it makes the DIS case more e+e-like, and does not introduce any additional parameters. But we have to accept that it is not the final answer, and that the model still has problems.
Mike Seymour