A reanalysis of titi (sooty shearwater, Puffinus griseus) banding data coll
ected between 1940 and 1957 by Lance Richdale demonstrates that well-docume
nted archival material can be usefully reanalyzed using newly developed sta
tistical techniques. In this study, we compare the results obtained by Rich
dale using empirical techniques to those obtained using a multistate mark-r
ecapture model. Although the two approaches produce similar estimates for s
ome of the parameters, the multistate model additionally provides estimates
of precision and can be used to answer biologically significant questions
not raised by the original worker. Our analysis provides some evidence for
two conclusions that Richdale put forward but could not justify rigorously:
(a) nonbreeders have a lower survival rate than breeders, although the dif
ference is not statistically significantly; and (b) nonbreeders are more di
fficult to capture than breeders. We argue that reanalysis should be carrie
d out more frequently on historical data and lament the fact that it can be
rare for such data to be made available for future scrutiny.