Mail*Link(r) SMTP               Historic Lifespans
 
I'd say the census data would be the obvious place to start but this would
give you the raw data. The population statistics are often analyised by
economic historians to work out mortality tables there is lots of this work
done in the Uk and Australia but I dont recall any specific US references
although, no doubt  there is some.
 
Iain Stuart
Uni of sydney