When using the pre-dose technique in ceramic dating, it is essential to add a calibration dose and to eliminate the background dose. For doing this measurement the sample requires multiple activation following multiple administration of a test dose to induce simulated "ancient" thermoluminescence (TL). However, the procedures introduce many factors that are difficultly corrected. This paper compares existing two kinds of measurement procedures and proposes a new procedure involving single activation following a single test dose administration. An experiment using the single activation procedure is carried out with natural fine quartz where the results obtained exhibit only 0.07 relative error. It appears that this new procedure increases the precision of the ancient TL measurement by about 0.04 and 0.11 over those of existing two kinds of procedures. In addition this method can give realistic values in TL dating for relatively "young" samples aged tens to hundreds of years where the standard high temperature TL method is relatively insensitive.