In 1849, Robert Mallet gathered all the necessary components to perform his experiment to record ground motion. Nevertheless, he failed to record adequate data (see Chapter 1). Today, the goals of seismic experiments certainly are more ambitious, but possible failures can arise from the very same causes. The causes of Mallet's failure are obvious:

• weak source: Gunpowder is 50% less energetic than dynamite.

• low sensitivity of receiver: It takes a fairly large ground motion to create ripples that can be observed at the surface of a mercury bowl.

• high instrument noise: Poor firing-time control, limited clock accuracy, and observer reflex time add up to constitute a large imprecision (noise) in time measurement.

• lack of understanding of wave propagation: A phenomenon as complex as wave propagation cannot be understood from one single observation.

In this chapter, I shall review briefly how the first three causes of Mallet's failure have been addressed in the past and shall discuss the way they are addressed today. I will try to refrain from predicting too much about how they could be addressed tomorrow. In Chapter 4, I will discuss the last cause of Mallet's failure — lack of understanding of wave propagation.