rats are on a normal calcium diet.” The comparison was based on water fluoride concentrations.

Several longer-term animal experiments are compared in Table D-1. The National Toxicology Program (NTP) (Bucher et al. 1991) and Maurer et al. (1990) experiments are well-known long-term fluoride carcinogenicity assays. Of the four studies, Maurer et al. (1990) added fluoride to feed; the others added fluoride to water. Figure D-1 shows results for male rats for the three studies that added fluoride to water. Fluoride bone concentrations for female rats were somewhat higher in the NTP study and somewhat lower in the Maurer et al. study. Femur and vertebra fluoride concentrations were similar in the Dunipace et al. (1995) study. Femur diaphysis fluoride concentrations were similar to concentrations in other sites, except for femur epiphysis, which was higher (Whitford and Birdsong-Whitford 2000; G. Whitford, University of Georgia, personal communication, June 2, 2004). Figure D-1 also shows regression lines through each set of rat data, as well as the crude and adjusted estimates for the human data (Zipkin et al. 1958) discussed earlier. The adjusted line estimates bone concentrations in males with 70 years of residence, but the slope is very similar to the crude model.

Assuming that linear models are realistic in this range and that rats at 18 to 24 months are roughly physiologically comparable to humans at 70 years (Dunipace et al. 1995), the committee compared the slopes for the human and rat studies. The estimates in the left column of Table D-2 (bone versus water) were computed by dividing the slopes for the human data by the slopes estimated for the Dunipace and NTP rat studies. (The commit-