What was done
The author used daily minimum air temperature data for the period 1906-1997 obtained from 36 U.S. Historical Climatology Network stations in the state of Illinois to calculate the date of last spring freeze, the date of first fall freeze, and the resulting length of the freeze-free growing season.

What was learned
In the words of the author, "(1) the date of the last spring freeze is nearly one week earlier now than it was 100 years ago, (2) fall freeze dates have not changed in a systematic fashion, and (3) the growing season is nearly one week longer now," which directly follows from observations 1 and 2.

With respect to the first of these observations, the author notes that it is driven by the century-long amelioration of the very coldest spring minimum temperatures and not by a uniform upward shift (warming) of the entire distribution of all minimum temperatures for the month in which they normally occur, i.e., April.  Likewise, he notes that the second phenomenon is a result of the fact that the very coldest autumn minimum temperatures have not changed all that much over the century of record, in spite of the fact that the entire distribution of all minimum temperatures for the month in which they normally occur, i.e., October, actually cooled at a very significant rate.

What it means
The complexity of the results makes it difficult to point to a particular forcing mechanism that could be responsible for them.  In terms of their impact, however, there is little question; they should be beneficial for agriculture, which is a blessing to all of us.