I always want to start out by pointing out that two closeby towns should, by all considerations, have very similar temperature records. But, as I have found over the past few months, this is not the case. I am taking a second look at Columbus and Seymour, Indiana tonight. Tonight I am looking in detail at the bias between the two towns.
Below is a chart of the temperature difference between Columbus and Seymour (Columbus minus Seymour) for the years 1992-1994.
You can see both wild differences in temperature between the two towns, but you can also see that for the 3 years shown here, in general, Columbus is much hotter than Seymour--by 3 degrees.
Let's look at 1992-1993.
You can see that in the winter of 1992-1993 the bias decreased a bit and then rose again in the spring.
1992 looks like this:
Now, all the above biases are positive. During 1971-1972 the bias was the other direction.
One sees a long stretch of negative biases from 1958 to 1964. Such things do not give great confidence that the temperature is being measured properly.