Toward the end of my discussion of observations, I ask whether simple observations, even with fancy equipment, is an efficient way to find out about the world. If what you're trying to observe is rare, then waiting for it to happen might mean waiting for a very long time. I talked about some of the ways that scientists can make their observations more efficient, but there are always going to be intrinsic limitations when one's methods are confined to simple observations. Think about it in terms of Mill's methods. If there's a combination of variables that we want to consider, but we haven't observed them and we don't know where to observe them, we're going to have to wait a long time, maybe even a lifetime in order to get these observations. But what if we could produce the combinations of variables that we were most interested in? Experiments are what allow scientists to do this. They allow scientists to generate the conditions that they want to observe by intervening on phenomena. There are at least two reasons why experimental interventions are so important. The first is that scientists use experiments to create new phenomena, and the second is to use experiments to isolate existing phenomena from the rest of the world. So in the first case, scientists sometimes literally create a phenomenon that can bring into existence a constellation of things that they wish to study. So this can be done, lets say, by generating a mutation, or making a new molecule. In some cases, the novel phenomena does exist somewhere in nature, but it's so rare, it's hard to find. But there are cases, such as when scientists synthesize completely new molecules that something genuinely new, genuinely new to the universe is created. The second kind of case is when we use experiments to isolate. This allows observations of nature to take place in a highly controlled manner. Isolating the phenomenon of interests for many other factors, especially ones it commonly co-occurs with. Now, this is a very abstract characterization of experiments. To get clear on it, let's visit a modern evolutionary genetics laboratory here at the University of Pennsylvania. How do we make these flies that encode the wrong species version of only a single gene? What we do is inject into an embryo three different components. One is called Cas9, a protein that's effectively molecular scissors, that we're going use to cut out the native version of this gene. Second, we have RNA's that are cognate, or they match the sequence of the gene that we're trying to target. These RNAs guide the Cas9 molecular scissors to the site we're trying to manipulate. Finally, we have a third component, which is DNA derived from the wrong species, that is used effectively as a template for the repair process. So the molecular scissors go in and cut out the gene of interest, and this third component provides a template from the wrong species that's then copied in to that genome. In the end, we're left with a fly that has the contemporary genes and telomeres of Drosophila melanogaster, but only at this one gene, the wrong species version. In this carefully engineered fly, we're able to very precisely hone in on the biology that has changed as a consequence of the evolution since these two species split. We learned from Professor Levine a little more about what a cutting edge genetics laboratory looks like. In her laboratory, she studies the evolution of the proteins attached to the end of chromosomes. In doing so, she illustrates the two reasons for turning to experiments that I mentioned earlier. The first is isolation. She isolates a specific phenomenon in a model species. She's looking at the behavior of proteins on the end of laboratory fruit flies chromosomes. She also creates a phenomena. She creates new genetic variations to study by using the technique of CRISPR Cas9, which lets her alter the DNA sequence of the fruit flies. This is of course just one laboratory, but we'd see the same kinds of motivations for experimentation in labs throughout the world.