Induction is the form of reasoning that allows us to move from the particular to the general, or from facts to laws. Laws state causal relationships between two or more observed facts. So there are two steps to every induction:

1. We seek out a causal relationship between two facts.

2. This relationship identified, we extend it to all empirical cases where it might apply.

Here's an example of an induction where these two steps can easily be distinguished: Pascal wanted to determine the cause of fluctuations in the column of mercury in a barometric tube.

First Step. Pascal noted that, in a certain number of cases, the cause of the fluctuations is the weight of the air. In other words, he discovered a law that governs the phenomenon in the cases he's observed. A causal relationship has been established.

Second Step. This relationship – which has been observed in a certain number of cases – is then extended to all possible cases, and Pascal asserts the general claim that the cause of variation in the height of the barometric column is variation in the weight of the atmosphere.

In the first step, Pascal sought to identify a causal relationship. How can such relationships be determined?

Mill, in his Logic, gave four methods for doing so – concordance, difference, concomitant variation, and residues.