One of the ways I like to rank months is by their anomaly from the rolling 30y mean.
First I calculate each month's anomaly from the previous 30 year mean (e.g for 2007 I would subtract it's value from the 1977-2006 mean).
I then find the standard deviation of the entire anomaly data set and then divide each anomaly by said standard deviation.
You can then rank each month by how exceptional it was compared the climatic norms of it time.
So for example, ranked in this way, December 1981(0.3C, -2.7 standard deviations from 30y mean) was colder relative to it's previous 30 years than was 1874 (-0.2C, -2.6 sd. from 30y mean).
Obviously this month will be by far the coldest when ranked this way and is likely to be the first December that hits -3 standard deviations. This is something you would only expect to happen less than once in every 200 years.