On the problem of obtaining reproducable thermoluminescence patterns, with special reference to alkali halides