Why do scientists measure?


John Oversby for the UK HIPST project: October 2009

Introduction


At first sight, like many philosophical questions, the answer to the question seems obvious. Measuring is what scientists do, and that is the end of it. We might also say that scientists measure because they want to quantify. The Cambridge Dictionary says quantify means 'to measure or judge the size or amount of something' so it looks like we have just used another word for measure rather than make things clearer! We might say 'to compare'. The Cambridge Dictionary says compare means 'to look for differences or similarities between two or more things'. This does not mean we have to measure. For example, we might say that one ball is red and one is green, and we would not be measuring. We might say that it is to be more accurate and the Dictionary says this means 'exact or correct, without any mistakes'. What started out as a simple question seems to be much more complicated when we look into it. We might, though, start with what kinds of things we measure by looking at the idea of 'properties'.

Properties


If you want to see an expert view of properties, then you can look at the Stanford Encyclopedia of Philosophy which has an article by Chris Swoyer, written in 2000. This is a simplified version for the HIPST project. Properties are attributes or qualities or features or characteristics of things. It seems that all these words mean something that is noticeable that describes a thing! It can be measurable, like the size, or how heavy it is, or how hot it is, or it may not be, such as its smell, or even who it belongs to. The properties we are most interested in can apply to many examples, and not just to one. Some properties can be related. We can say that one thing is larger than another, or that it has a stronger smell, or that it is hotter than it was. These are comparisons and are mainly about differences or similarities. Some of these comparisons are subjective, that is they depend on who is making the comparison. 'Stronger smell' is one of these and different people may make different judgments about whether one thing has a stronger smell than another! These kinds of judgments have a place. In hospital, patients are often asked to give a number from 1 to 10 to describe their pain. What is most important here is to have an idea about how much stress the patient has from the pain, not whether it is exact. In other cases, scientists wish to repeat their experiments, or to repeat the experiments of others. In these cases, it is helpful to have a measurement. In other words, the property measurement is there to solve a scientific problem, of repeating the experiment as close to an exact copy of the first time as possible. Properties, in general, exist to help us solve the problem of describing things, usually by using general properties that many things can have. The concept of properties exist to help in creating explanations.

Measurement


Measurement seems to be used by scientists for a number of purposes:
1. to support faithful replication of work by others;
2. to enable a detailed analysis of data, often based on statistical methods or in testing theoretical ideas;
3. to characterise and elaborate a concept;
4. to provide for further predictions, known as fertility;
5. to examine the smoothness of data, that is, whether there are some otherwise hidden surprises in the collected data;
6. to promote the development of instruments for exploration.

Measurement is beset by various issues such as:
1. Validity, or does the measurement measure what it claims to do?
2. Reliability, or is the measurement repeatable?
3. Precision, or is the error in measurement so small that the value is very close to the actual value?
4. Accuracy, or is the error in the measurement so biased that even an average is some way from the actual value?

In the following description, I will use temperature as my measuring example.

I place validity at the top, since improving the other three are no good if the scientist is not measuring what she claims to be measuring. Some properties, such as temperature, are almost impossible to measure directly. Scientists have to make do with measuring another property that changes with temperature, such as the length of a liquid in a glass tube against a scale, or the pressure of a gas in a container. They face a big question about whether they are really measuring temperature. This question will not be extended in this short article. Measurement of temperature takes some steps:

A. Choose some property that changes when the temperature changes. It is best to choose a property that gives only one value for each temperature. For example, suppose a pressure difference of 2 units was found and measured at two temperatures. On another occasion, another person measuring two units would not know which one of the two temperatures was being measured. Fortunately, the length of a liquid in a glass tube against a scale, or the pressure of a gas in a container are two good properties for this.

B. Choose some method of standardisation. For normal thermometers, scientists have chosen the temperature of a mixture of ice and water as one temperature (fixed point) and the temperature of a mixture of steam and water at standard atmospheric pressure as the other temperature. Scientists have also carefully made clear the conditions so that the standardisation is very well repeatable in different laboratories. For example, many liquid in glass thermometers have '76 mm immersion' engraved on the stem. This means that the thermometer has to be 76 mm inside the ice/water or water/stem mixture to be sure it is reading correctly.

C. Now the thermometer is left at each standardising temperature until it stops changing (do not forget the 76 mm immersion). This might take some time, since different thermometers take different times, and is typically a few minutes for school laboratory thermometers, but can be less than a second for some electronic thermometers. An engraved mark is made on the glass to show the length of the liquid at the fixed point. For a gas thermometer, a mark can be made on the pressure gauge.

D. The space between the two marks is divided into 100 with a scale. The scale can be used when using the thermometer in the future. Strictly, the thermometer is only correct at the two standardising temperatures, but is used at the other ones, too.