I did the same analysis as nei in #2 for Berlin-Tempelhof and Frankfurt from 1949 to 2012. I used "R", don't know how to visualize it better cause I'm a beginner in statistics.
You can notice that July has more positive outliers and January more negative outliers from the mean in both cities and the deviations are generally higher in January.
1. January deviations from the mean in °C:
Results:
Frankfurt Estimate: 1.059e+00
Frankfurt Pr(>|t|): <2e-16 ***
Adjusted R-squared: 0.8421
There is a significiant relationship between Frankfurt's and Berlin's January differences from the mean. But I have got a problem with Heteroscedasticity, maybe an other formula than y=a+bx+u is needed.
2. July deviations from the mean in °C:
Results:
Frankfurt Estimate: 9.189e-01
Frankfurt Pr(>|t|): <2e-16 ***
Adjusted R-squared: 0.8285
There is also a significiant relationship between Frankfurt's and Berlin's July differences from the mean. No problems with Heteroscedasticity etc. detected.