No Gordon, you were not far from clear. You were perfectly clear . Rereading your earlier post, I realize that I had just not read it properly. The problem is mine, not yours (Napoleon used to tell his assistants: "Go slowly and carefully because I'm in a hurry).
OK, so you mean with your modification, the gap is as long as the longest possible answer for THAT particular question.
So, yes, I was thinking exactly along the lines of what Tom said: this is not desirable because for Cloze type exercises many times there is only one answer possible and as Tom says, the length is very often a give away.
What I don't understand, though, is why you say that doing it the way Tom suggests would be a departure from the way that JCloze currently works. If I'm not mistaken (and given the history of my previous postings, I probably am ) Hot Potatoes JCloze already generates gaps that are longer than the correct answers for a particular question. As I said in my initial posting, when I generatate an HTML file from HP JCloze, the gaps ARE most of the time longer than the correct answers (I asssumed that they were as long as the longest answer for the whole exercise). It was only when I uploaded the files in the HP format that I saw that the gaps were short.