Some scientists, especially astronomers, observe; this is the most difficult skill in science. It is only very rarely possible to know exactly what it is you’re observing; and when you do, by definition almost, you can’t find anything interesting, surprising or new. Sampling what you observe, or choosing a technique for examining it, has built-in problems that we can list, but which usually can’t be avoided altogether even if you’ve identified them. We will discuss the relationship of observational uncertainty to statistical methods in Chapter 6. When critical observation is necessary, it is always as well to bear in mind one of the precepts of Rabbi Akiva: ‘Take the long route which is shorter, not the short way which turns out to be much longer….’ It is worth spending a long time making sure that you are familiar with what you are going to observe; you should be able to observe what other people have seen using those methods, even if they were mistaken! Ideally, you should be a good enough observer that you can tell why past observers were mistaken. Schiaparelli saw those ‘canali’ on Mars; any mottled pattern can produce the illusion of a network of lines, so the problem was to determine if the lines (exciting) or the mottled pattern (boring) were 38what you should observe. It will also be better science if you can use the apparatus at least as competently as your predecessors, and adjust and repair it if necessary. Like the apocryphal American ultraviolet light sensors, automated sampling substitutes other people’s prejudices for yours, making it much more difficult to locate them. Only if you know the material thoroughly, and both the advantages and demerits of your particular piece of apparatus or software can you recognise unavoidable, built-in problems. So make sure that when you use the apparatus to observe ‘standard’ material you can reliably and repeatedly get the ‘standard’ answers. Only then will people believe your surprising, mind-changing observations.