Returns in the 70s weren't quite as bad as the DJIA graph suggests, because the companies in that average would have been paying substantial dividends.
I thought that the DJIA divisor was adjusted for dividends so that the pre-dividend and post-dividend index values were the same (essentially, acting as if investors took any dividend and re-invested it by buying shares of the DJIA constituent companies).
It sounds like they only consider stock dividends that are effectively a form of split.
Anyone has the article he's talking about? Scribd won't let me to consume it.