It was last field season and U.S. Geological Survey hydrologist Peter Spatz and I were making active layer measurements by a stream. In polar regions, the active layer is the uppermost layer of soils that thaws and freezes seasonally. It lies over permafrost, which is soil or rock that remains below zero degrees Celsius for two or more consecutive years. A simple way to measure the depth of the thawed active layer is to use a metal, T-shaped rod, push it into the ground, and use centimeter markings on the rod to determine how far it goes in before it reaches the frozen boundary.

Pete, Karen, and Mr. T

That's what Pete and I were doing, pushing Mr. T into the ground. "Come on Karen!" Pete encouraged. "Put some chili bean on it." Then Pete looked at me, shook his head, and said, "You really ain't got none." In stream team vocab, chili bean refers to body mass, the kind that you can throw onto a shovel or plop down onto something like a T-rod and let gravity do its thing. As the smallest stream team member both last year and this, though, I don't have a lot of chili bean to spare. And that made me wonder.

In most cases I thought I knew when I reached the frozen boundary. To me, it felt pretty much like I was hitting concrete, but could I be sure? Were there errors in my measurements? Does chili bean have an effect?

Experiments are generally considered to have two types of errors - random errors and systematic errors. Random errors may be due to human mistakes like misreading Mr. T or, in some cases, to sampling. For instance, because of natural environmental variability, it's possible that the active layer depth in the spot we've chosen to take the measurement may not be representative of the surrounding active layer depths. Random errors are just as likely to be positive as they are to be negative, and they're not repeatable.

A chili bean effect though would more likely lead to a systematic error. Systematic errors can result from faulty equipment or technique, and they occur in a particular direction. If someone with more chili bean could get the T-rod to go deeper into the ground than me, then my measurements would tend to be less than the actual values. While random errors, being random, may cancel each other out, systematic errors may result in a bias.