Philosophy Lexicon of Arguments

Experiment: artificial bringing about of an event or artificial creation of a state for testing a hypothesis. Experiments can lead to the reformulation of the initial hypotheses and the reformulation of theories. See also theories, measuring, science, hypotheses, Bayesianism, confirmation, events, paradigm change, reference systems.
Author Item Excerpt Meta data
Duhem, Pierre
Books on Amazon
Experiments I XIII
Experiment/Duhem: The experimenter does not merely read the instruments, he uses the instruments. The following are required:
1.Theory about the domain in which measurements are made.
2.Theory of used measuring instruments. In every device there are systematic errors, the experimenter knows that. Duhem: the Ex works constantly with two devices: the concrete measuring device and an ideal structure for which the physical laws "exactly" apply.
Between the phenomena ascertained in the experiment and the result of the experiment formulated by the physicist is a complex translation process. The facts which the experimenter determines are already "theoretical facts". The experimenter must always use theories for his work. The results are only useful for other physicists, if it is know which theories he has used. And whether corrections were already made.
An experimentum crucis (whose failure would disprove the whole theory) is rejected. (> Holism).
VsExperimentum crucis: (for example, by Quine, 1951, as the main attack against logical empiricism: Holistic view of science.) No experiment can show where the fault is located in the system. The examination of a particular hypothesis is only possible by using a whole group of further laws, ultimately the whole theory.
Since Bacon, hope was linked to an experimentum crucis: a decision-making experiment between competing theories.
E.g. Foucault's attempt to decide between Newton's theory of emission of light and Huygens' undulation theory sees Duhem as an experiment of the experimentum crucis. Foucault was able to show that light actually spreads slower in water, which made Newton look refuted. Duhem shows that this conclusion is not valid. The error could be in a secondary hypothesis.
I 189
Experiment/Duhem: disintegrates into two parts.
I 190
1. Observation of certain facts. For this, the knowledge of physics is not necessary at all.
2. Interpretation of the observed facts. For this you have to know the physics. The physicist does not write in his protocol that the instrument has indicated a certain stroke, he notes that the gas has reached a certain volume.
What are the size of the volume, the temperature, the pressure? Three concrete objects? No, there are three abstract symbols that connect physical theory with facts.
I 192
Definition experiment: accurate observation of a group of phenomena, plus interpretation of them.
This interpretation replaces the concrete given with the help of the observation of actual ones obtained by abstract and symbolic representations which agree with them on the basis of the theories which the observer accepts as permissible.
I 193
Experiment: The result is the establishment of a relationship between different concrete facts.
A certain fact was artificially produced, another stems from it.
I 210
Experiment/Duhem: We must know the theories that the physicist considers valid.
During the experiment, the physicist continued to operate two apparatuses side by side: the actual measuring apparatus and the ideal one defined by mathematical formulas and symbols.
I 215
The result of a physical experiment has not a security of the same level as a fact, established by non-scientific methods, from a healthy person by simple observation. The experiment is less direct and more dependent on a whole group of theories.
I 273
Experiment/Duhem: The experiment is not the basis of theory, but the highlight. The totality of the theorems gives an ever more similar picture of the totality of the experimental facts.

Duh I
P. Duhem
Ziel und Struktur der physikalischen Theorien Hamburg 1998

> Counter arguments against Duhem
> Counter arguments in relation to Experiments

back to list view | > Suggest your own contribution | > Suggest a correction
Ed. Martin Schulz, access date 2017-03-30