1. No, there would be no valid evidence against the law unless many people (hopefully some of them a bit more experienced than high school students) using many different experimental setups all found that energy was not conserved. One cannot disprove a hugely well-established law with one experimental result--especially when that experimental result comes from a high school lab, where margins of error tend to be enormous.
2. Experimenters manipulate only one variable per experiment to ensure that the results they observe are actually due to that variable. (“Ensure” might be too strong a word here, because it is always possible that the experimenter forgot to account for a variable that is actually important. However, assuming an ideal experiment in which the experimenter correctly identifies all the variables that could significantly influence the results, manipulating only one variable ensures that the results are due to that variable.) If experimenters manipulated two variables at once, they would have no idea which manipulated variable--or what combination of the variables--created the results.
3. It would be logical to control reactant type, the shape of container used, and the initial volume of each reactant (10 mL of low-concentration reactant for the first trial, 10 mL of highly concentrated reactant for the second trial, etc.). It is possible, although unlikely, that factors such as room temperature or barometric pressure, the manufacturer of the reactants, or the age of the reactants might also affect the results. (Extreme temperature differences would affect reaction rate, but probably the relatively minor differences in room temperature from one day to the next would not have much of an effect.) It would be illogical to control eye color of the experimenter, brand of light bulb illuminating the room, or the tide at the time the experiment takes place.
I am able to make these distinctions because of my prior knowledge about reactions and the factors that affect them. I have never heard of a scientist controlling for eye color in this type of experiment, and everything I know about eye color and reactions rate implies that the two are unrelated. Therefore, I can safely assume that eye color does not need to be controlled.2 Of course, this approach is not without risk. Interesting discoveries often result from the realization that variables no one thought needed to be controlled actually influence the results.
4. Theoretically, absolute laws of history and sociology probably exist. If a computer had infinite knowledge about all the people within a society and infinite knowledge of history, it could probably make accurate predictions about people’s future actions.3 However, realistically speaking, universal laws are impossible to discern, because it is simply too complicated to account for all the variables involved in the interactions of large groups of people. However, statistical laws may be achievable, and it is certainly possible to recognize trends in human behavior. If general laws of human behavior could not be discerned, everything from writing effective laws to designing manipulative advertisements would be much harder. While policy makers, political analysts, economists, and marketers are not completely accurate in their predictions, they are often at least somewhat accurate, implying that we do have some knowledge of the laws of human interactions.
5. Human observation alone is not very exact, especially on such a small scale as the expansion of a metal bar during heating. If I was even able to detect the metal bar’s expansion without measuring the length numerically, I would be unable to say anything more than, “It gets a bit longer as it gets hotter.” This may be an interesting observation, but it is inexact. Qualitative observation works fine for expressing the general trend, but it can get no more specific than that. Without numbers, it is nearly impossible to determine whether length and temperature are related linearly or with some more complex relationship. Because of this, the general trend is of limited utility, because it cannot predict the length of the bar after heating or cooling beyond general statements such as, “It will probably get a little longer.” If all science had to be expressed in such hazy terms, it would be hard to engineer objects of any complexity, because the interactions of the components would be too unpredictable, not to mention the difficulty of building anything without numeric measurements.
1. And yet, wouldn’t you only need to see one black swan to uproot any belief that there exist only white swans?
2. There’s a kind of circularity here – I know what factors influence the experiment because I know what factors influence the experiment.
3. But is this possible? For a small, closed system, for example, it’s one thing for a computer to have knowledge of every particle at every time, but it’s much more for the computer to extract laws of behavior from that knowledge.