I was looking back at the 2000 election based on some comments in another diary and stumbled across Gallup's polls back to 1936.
While it struck me that 3rd parties have played interesting spoiler roles more than I thought (Who was that Wallace guy in 1968 who got 15% of the vote anyway), what really got me thinking was the 1936 and 1940 trendlines. 1936 was actually more like Clinton's 1996 win, but 1940 looks a lot like this one so far.
Between July and Sep, they were both deadlocked in the mid 40s, then there was a convention bounce for Roosevelt which held up.
So far, that is what this year looks like too, at least on the presidential front. The overall political situation isn't all that different either, although the middle east unrest and Afghanistan isn't nearly as strong as the wars raging in Europe and China at the time, but from a domestic standpoint, Roosevelt had a hostile supreme court, had endured misguided austerity policies and a slow but sputtering recovery.
Look at the 1940 graph here http://www.uselectionatlas.org/...
and compare the July-Sep info on the graph here: http://www.gallup.com/...
You can quibble about Gallup's voter screen but these at least are likely apples-to-apples comparisons in that they're both registered votors and the same polling company (if separated by over 80 years in time). One thing to note is that while FDR's bounce was in a similar range, his opponent dropped to 40% for a while, before coming home a lot closer in the final stretch. Romney's floor seems a bit higher.