This, however, is not what I mean to imply by the term empirical observation. I mean rather thinking on the basis merely of facts which occur in the natural course of events, which have not been systematically produced by ourselves or others for the purpose of solving a problem. Logicians usually call this method simply _observation_, and oppose it to experiment. But I object to calling this simply observation because experiment itself is really observation, only in one case we observe merely events which happen to occur, and in the other we observe the results of events _which we have made occur_. The true way of distinguishing these two methods would be to call one _empirical observation_, and the other _experimental observation_. This empirical method—if indeed I am justified in calling it a _method_—is the most common in all thinking. To give examples of it would be to show how men generally think. But the method has real value, and may even be the most important of all, for if we thought without it our ideas would doubtless be original, but very dangerous. Let us apply it to some of the problems considered under other methods. Empirical observation is used where experiment is impossible—often, unfortunately, where experiment is merely inconvenient. In political science the empirical method would consist in noting the effect of certain laws,—e.g., tariffs of different countries and of the same country at different periods—and noting economic conditions at the time the different tariffs were in effect. Allowance would be made for other factors which could influence the country’s economic condition and the effect of the tariff could then be determined. The empirical method of dealing with meteorology, the science of weather, would consist in making a study of cloud formations, wind velocity, moisture in the air, temperature, etc., and noting what conditions usually or perhaps invariably followed certain of these conditions. From this, conclusions could be drawn as to what weather to expect following certain conditions. But valuable as empirical observation is, and often as we must use it, it should never be employed when we can experiment. When the empirical method is rightly used allowance always has to be made for certain irrelevant factors. But “making allowances” is always sheer guess work. _The experimental method consists not in making allowances for certain factors, but in eliminating those factors._ In our example from political science experiment is practically impossible, because the factors which may influence economic conditions are innumerable, and even were they few, no country could survive the dangers of being experimented upon—to say nothing of its permitting it. Experiment is similarly impossible in dealing with weather conditions directly. It is impossible in astronomy.