We're not talking about counting pets from the last rerelease, this project is for 2008 so it is not time-dependent.
Regarding counting all 2008 pets, there were 922987 pets adopted in 2008, the last of which was this one - http://www.chickensmoothie.com/viewpet.php?id=922987
That's too many to count? There were only about 20k pets adopted during Dec 18 in 2008 so we could always do a count of the rerelease; with that said I'm not sure what normal adopt pets may also have been picked up then which would skew the data.
If you guys want, we could start with July 2008 since we do have some July data already, to validate that selecting a sample to count is a valid method?
One issue we will encounter with August 2008 is that there were no adoption limits, so if people went crazy over balloons but didn't pick up many tribals, balloons will appear to be more common. But this is not actually correct because the rereleases don't consider adoption rates, they only consider rarity ratios inside litters. Perhaps counting the 2008 rerelease or the 2009 rerelease would give us better data? Alternatively, we could take the amount of individual pets adopted from each litter into consideration when determining their ratios.
Back in 2008, if you adopted 50 tribals and 30 balloon dogs, you did that because that is how many you wanted, not because of rarity differences or adoption differences. But if were to compare the rarities of say the first 30 tribals to the 30 balloon dogs, we could get an idea of interlitter ratios? Is that valid?