Standardized mortality ratio

Why is it that when computing an SMR in an age stratified population adding up the expected number of deaths in each age group yields a slightly different number then when doing the calculation just the total? Is it because of rounding differences?