40 



A. Comfort 



roughly X 3-2. The last part of the curve calculated from q^ 

 values at ages of 25 and over is, of course, largely arbitrary, 

 and in the late intervals losses, many of them from age- 

 dependent causes, equal or exceed deaths. 



The survival curves and expectations of the separate cohorts 

 are roughly similar, but there is substantial scatter (Table II), 



lOOi- 



SURVIVAL CURVES OF THOROUGHBRED 

 MARES -GENERAL STUD-BOOK 



AGE IN YEARS 



Fig. 1. Survival curves of mares foaled in 1860-64 (- 



-), in 1875-80 



( ), of brood mares by three selected long-lived stallions (Hampton, 



Galopin and Hermit) ( ), and of grey mares ( ). 



the last three years of the 1875-80 sample having a shorter 

 expectation of life than any of the others (0-05 < P < 0-02 

 for the largest difference). The proportion of scoreable lives 

 to total fillies foaled is rather higher in sample B (33-0 against 

 26-8 per cent) but there is no immediately obvious reason 

 for the differences in cohort performance. There may be bias 

 in sampling, since the two lowest-scoring cohorts of sample A 

 contain an unusually low proportion of late entrants to the 

 record. Mares are normally entered when first covered, but 

 animals among the unaccounted majority which have not 



