We re-analyse the data of Yamada et al. (1995b) which represent the fi
rst substantial published experiments on thermal annealing of fission
tracks in zircon, We use an improved approach to the statistical model
ling of the relation between track length, temperature and time that h
as recently been developed for fission tracks in apatite. There are se
veral advantages to our new approach: we distinguish between unanneale
d and induced tracks; we partition the residual Variance into ''within
'' and ''between'' components corresponding to sampling error and extr
aneous sources of variation; we allow for different numbers of tracks
in the experiments; and we use the information even in those cases whe
re no tracks are seen. Our conclusions differ somewhat from those of Y
amada et al. (1995b) in that we find that the data are not extensive e
nough to distinguish clearly between competing models. There is a need
for further experiments to be planned and carried out, and we make so
me recommendations on their design. (C) 1997 Elsevier Science B.V.