“If births were evenly spread across quarters, then #10640 would not be that far off..but I cannot find a periodic breakdown for Hawaii in 1961.”
Birth distribution by month for US in 1961 is here:
http://www.cdc.gov/nchs/data/vsus/vsus_1961_1.pdf (p. 1-29)
Birth distribution by month for HI in 1961 is on p. 2-6.
Check my math, but I get 9942 live births recorded through July. There were 1460 in August. I believe Obama BC was not registered until the 8th, so 1460 x 8/31 = 377. Add the latter to 9942 and you get 10,319 vs. Obama’s actual number of 10,640. Now, you could say Obama’s record “should” have been recorded on 8/14 given its sequence number (i.e., 31 x (10,640 - 9942)/1460 = 14.8). But births obviously are not spaced exactly evenly across every day of the month. As one example, for all we know, there may have been a “bumper crop” of Kennedy-election-night babies born in Hawaii in early August, in which case Obama fits right a mini-baby-boom in the first week of August. Indeed, this might explain the delay of several days between birth and actual registration: an unexpected, but quite temporary overload for the staff assigned to do this etc.
I think this reasonably suggests whatever other things are suspicious about Obama’s certificate, late registration isn’t one of them.
Actually, using an average for July-Sept would yield a more valid monthly estimate, when divided by 31, yields 48 per day. I don't necessarily buy the "bumper crop" argument as the total for August was 95% of July and 92% of September. I'm going to go with 48 per day as a slightly liberal estimate.
So, 8 * 48 = 384 + 9942 = 10,326, or 314 births short of 10,640...or about 8 days later than Aug 8th.
Now, that's a lot of wiggle room to find someone who was born in August 1961 but no longer living on August 20, 2008, the day before Factcheck posted that Cert. number.