1.  First results of this research were presented at the 1998 Canadian Population Society Annual Meeting [26]; more detailed results are available in [25]. 
2.  The hypothesis about equal distribution of deaths between the younger and the older cohort is a simplification. In fact, because of rapidly declining numbers of persons with age, combined with a slow increase in the death rate, we would expect more deaths in the younger cohort (lower triangle of the Lexis diagram). However, there is a compensating effect because the upper triangle (older cohort) contains more winter months, and thus deaths at older ages tend to be distributed almost equally between the two cohorts. 
3.  The exceptional case of Jeanne Calment, a French woman who died in 1997 at the age of 122 years and 5 months is not mentioned in this paper. Other validated cases of exceptional longevity are also omitted: Marie Louise Meilleur, a French Canadian woman who died in 1998 at age 117 and Christian Mortensen, an American man who died in 1998 at age 115. 
4.  According to the census figures, Canada had 3685 centenarians in 1991: 135 centenarians per million of the total population. 
5.  To reduce annual fluctuations, death rates were smoothed twice through a moving average, over three and five years, between 195559 and 19851989. To cancel the effect of structural variations in calculating the rates, Sweden’s population distribution by fiveyear age groups was used for all countries (Sweden: 8084 years: 0,6360; 8589 years: 0,2743; 9094 years: 0,0774 and 9599 years: 0,0123). Given that fewer men die over the age of 80 years, rates for the total population are composed of twothirds of female rates and onethird of male rates. 
