In a laboratory, a biologist carefully counts out seeds and plants them in two containers of soil, one container held at a cooler temperature and one at a warmer temperature. She keeps them watered for several days and counts the seedlings when they appear. She calculates the fraction of seeds that actually sprouted (the germination rate) and notes that it is higher in warmer soil than in cool soil. She decides on a model that seeds sprout better in warm soil than cool soil. Based on this model, she predicts that in very cold soil no seeds will germinate and in very hot soil all seeds will germinate. To test the model and the prediction, she throws out the first batch of plants and runs the experiments again, this time using several containers of soil at several different temperatures. The results for cool and warm soil agree with the first experiment, confirming those results. No seeds sprout in the coldest soil, confirming that prediction. But in the hottest soil, no seeds sprouted at all! She modifies her model: germination rate increases with soil temperature, but at the hottest temperatures the seeds are “cooked” before they can sprout.
Experimental science is the primary type of science done in the fields of physics, chemistry, and molecular biology, as well as in certain aspects of ecology and geology. In the laboratory experiments are accessible; the scientist can measure what is happening, monitor the experiment from beginning to end, destroy the products of the experiment, and start over at any time. She can control many variables in the experiment (such as soil temperature) and remove external variables (such as rabbits eating the seedlings). And she can repeat experiments in the lab as necessary to confirm the first results. Basic assumptions, such as the underlying consistency of physical laws, can be tested by repeated experiments. Experimental scientists make testable predictions (such as the germination rate in very hot soil) that can be confirmed or contradicted in future experiments.
Another important tool of science is making careful observations. Sometimes controlled experiments cannot be done because the system under study won’t fit in the lab, is too far away, or is too dependent on its environment. In those cases scientists can still make careful observations. This method is also used when there are ethical reasons to avoid a full range of experiments, such as in medicine. Here’s an example of observational science:
An ecologist travels to the site where a forest fire occurred the previous year to study how the forest is recovering. He carefully counts all of the plant seedlings in a certain area and notes what types of plants they are. For the next ten years he returns once a year to count the growing plants. He finds that the wildflowers are the first to sprout and grow, but after some years tree seedlings are starting to compete with them. He hypothesizes that wildflowers grow better than tree seedlings in sunny, open spaces, but as the area gets crowded with plants and becomes shadier the tree seedlings do better. He predicts that this same recovery pattern will be observed at other forest fire sites and that the growth of tree seedlings will depend on the amount of shade in the area. He repeats the observations at another fire site where the fire was more widespread. Because fewer trees survived this fire, the site has less shade. He confirms his hypothesis that wildflowers are the first plants to appear but sees far fewer tree seedlings than at the first site. This corresponds with his model that shady conditions are important for the appearance of tree seedlings.
Observational science is commonly done in the fields of meteorology, ecology, medicine, astronomy, and geology. Typically the objects studied in observational science are less accessible than those in experimental science. The ecologist can’t sit all year and watch the plants grow, and an astronomer can’t travel to a star to measure its temperature. But scientists devise alternate methods to get around these difficulties, such as counting plants periodically or analyzing the light of the star to deduce its temperature. Observational science is not controlled; meteorologists cannot produce a cold front whenever they like, nor do ecologists burn down forests just so they can watch how they recover.
Observational science must take nature as it comes. A lack of control makes observational science less repeatable than experimental science. The forest fire can’t be repeated whenever the ecologist wants, but fires happen often enough that many sites are available to study. Usually enough examples are available that the consistency of the underlying laws of nature can be tested on several cases. Observational science is a useful partner to experimental science, because observations can be made in many situations where experiments cannot be done. But just like experimental science, observational science makes testable predictions (like the wildflower growth rate after a fire) that can be confirmed or contradicted in observations of other, similar systems.
A third method of scientific investigation is modeling the past behavior of systems, including events that occurred before they could be directly observed. Here’s an example:
An ecologist travels to a remote forest in order to study its history. She first examines a large tree that has recently fallen down in a storm. She takes a thin slice of the trunk back to the laboratory and counts the tree rings. She finds that a particular ring from 131 years ago is extremely thin (indicating drought) and shows evidence of mild fire damage. She hypothesizes that much of the surrounding forest burned down 131 years ago but that this tree survived. Based on the work of her colleague who studies recent forest fires, she makes predictions about the other trees living in the forest: the largest trees will show similar fire damage from 131 years ago; many of the other trees will prove to be 120-125 years old, having sprouted 5-10 years after the fire. To test this prediction, she takes core samples of several living trees and looks at their rings. The results confirm her prediction: the older trees all show evidences of fire damage from 131 years ago, and many of the other trees are about 120 years old.
Historical science is common in the fields of ecology, climatology, astronomy, cosmology, evolutionary biology, geology, and paleontology. The goal of historical science is to deduce the natural history of systems such as forests, rocks, and planets. Historical science is not directly accessible because no scientists were around at the time to make observations; however, those events are indirectly accessible because of the evidence they have left behind. Like a detective, a historical scientist uses the evidence available today to deduce the history.
Like observational science, historical science is not controlled: scientists cannot go back in time to change the initial event, so they have to work with what actually happened. Historical science investigations can be repeatable when many similar historical situations are available to study (such as the many different trees born after the same forest fire). In some cases, however, the event is not repeated (as in the case of the universe: there is only one universe for cosmologists to study), but scientists can still find evidence that tells them about the natural processes that occurred during that event. Historical science, at its best, is particularly useful for testing whether physical laws remain unchanged over the years, because historical science gathers data related to events that happened over as wide a period of time as possible.
Most important, historical science makes testable predictions, just as experimental and observational science do. Scientists routinely study one system (such as one tree or one star cluster), make a model for its history, and then predict what they will find in additional observations. These observations could be of other similar systems, or they could be of the same system but made with different instruments. In either case the observations test the prediction, supporting or contradicting their model for the history of the system.
All Three Methods Needed
These three styles of science blend into each other. As experiments become more complex and less controlled, they become more like observational science. Observational scientists often do experiments in labs to help them better understand what they are learning. The historical models used in historical science depend on observations of present data, and observations of present behavior are more useful when you have a good model of the past history of the system.
All three methods of gaining scientific knowledge are necessary. Systems such as forest recovery are too large and complex to be brought into the laboratory and must be studied observationally. But laboratory experiments about seed germination are helpful to foresters in understanding which plants are most likely to sprout after a fire. Historical studies of past fires help scientists understand how forests can recover over the long term. Each of these three types of science can make new discoveries that are then tested by the other types of science. These different styles of investigation reinforce and correct each other, leading to a better understanding of the natural world and its history.