How did Mendel's dihybrid crosses help him develop his second law?

In the dihybrid cross, you're looking at the probability for two different traits. For an example, let's use seed color and texture. For color, yellow is dominant over green, and for texture, round is dominant over wrinkled. When he crossed a pure yellow/round with a pure green wrinkled, all the offspring were yellow/round. When he crossed the offspring from the first cross, the results were plants that had seeds that were yellow/round, yellow/wrinkled, green/round, and green wrinkled.

So from this he was able to conclude that whatever caused the plants to inherit specific traits (remember, he didn't know about DNA, it hadn't been discovered yet!), it wasn't linked together - otherwise the offspring from the second cross would have reverted to the same two phenotypes as the parental generation - yellow/round and green wrinkled. The fact that plants with yellow seeds could now be wrinkled, and plants with green seeds could also be smooth, showed that these traits were inherited separately, and could combine randomly