From The Philosophy of Physical Science by Arthur Stanley Eddington: Chapter II Selective Subjectivism

Let us suppose that an ichthyologist is exploring the life of the ocean.  He casts a net into the water and brings up a fishy assortment.   Surveying his catch, he proceeds in the usual manner of a scientist to systematize what it reveals.  He arrives at two generalizations

(1)   No sea-creature is less than two inches long. 
(2)   All sea-creatures have gills.

These are both true of his catch, and he assumes tentatively that they will remain true however often he repeats it.

In applying this analogy,  the catch stands for the body of knowledge which constitutes physical science, and the net for the sensory and intellectual equipment which we use in obtaining it.  The casting of the net corresponds to observation;  for knowledge which has not been or could not be obtained by observation is not admitted into physical science.

An onlooker may object that the first generalization is wrong.  "There are plenty of sea-creatures under two inches long, only your net is not adapted to catch them."  The ichthyologist dismisses this objection contemptuously.  "Anything uncatchable by my net is ipso facto outside the scope of ichthyological knowledge, and is not part of the kingdom of fishes which has been defined as the theme of ichthyological knowledge.  In short, what my net can't catch isn't a fish."  Or- to translate the analogy - "If your are not simply guessing, you are claiming a knowledge of the physical universe discovered in some other way than by the methods of physical science  and admittedly unverifiable by such methods.  You are a metaphysician.  Bah!"  

The dispute arises, as many disputes do, because the protagonists are talking about different things. The onlooker has in mind an objective kingdom of fishes.  The ichthyologist is not concerned as to whether the fishes he is talking about form an objective or subjective class;  the property that matters is that they are catchable.  His generalization is perfectly true of the class of creatures he is talking about - a selected class perhaps, but he would not be interested in making generalizations about any other class.  Dropping analogy,  if we take observation as the basis of physical science, and insist that its assertions must be verifiable by observation, we impose a selective test on the knowledge which is admitted as physical.   The selection is subjective, because it depends on the sensory and intellectual equipment which is our means of acquiring observational knowledge.  It is to such subjectively-selected knowledge and to the universe which it is formulated to describe that the generalizations of physics - the so-called laws of nature - apply.  

It is only with the recent development of epistemological methods in physics that we have come to realize the far reaching effect of this subjective selection of its subject matter.  We may at first, like the onlooker, be inclined to think that physics has missed its way, and has not reached the purely objective world which, we take it for granted, it was trying to describe.  Its generalizations  if they refer to an objective world, are or may be rendered fallacious through the selection.  But that amounts to condemning observationally grounded science as a failure because a purely objective world is not reached by observation.  

Clearly an abandonment of the observational method of physical science is out of the question.  Observationally grounded science has been by no means a failure;  though we may have misunderstood the precise nature of its success.  Those who are dissatisfied with anything but a purely objective universe may turn to the metaphysicians, who are not cramped by the self-imposed ordinance that every assertion must be capable of submission to observation as the final Court of Appeal.  But we, as physicists, shall continue to study the universe revealed by observation and to make our generalizations about it; although we now know that the universe so reached cannot be wholly objective. Of course the great mass of physicists, who pay no attention to epistemology, would have gone on doing this in any case. 

Should we then ignore the onlooker with his suggestion of selection?  I think not;  though we cannot accept his remedy.  Suppose that a more tactful onlooker makes a rather different suggestion:  "I realize that you are right in refusing our friend's hypothesis of uncatchable fish, which cannot be verified by any tests you and I would consider valid.  By keeping to your own method of study,  you have reached a generalization of the highest importance - to fishmonger's who would not be interested in generalizations about uncatchable fish.  Since these generalizations are so important   I would like to help you.  You arrived at your generalizations by examining the net and the method of using it?"  

The first onlooker is a metaphysician who despises physics on account of its limitations; the second onlooker is an epistmologist who can help physics because of its limitations.  It is just because of the limited - some might say, the perverted - aim of physics that such help is possible.  The traditional method of systematic examinations of the data furnished by observation is not the only way of reaching the generalizations valued by physical science.  Some at least of these generalizations can also be found by examining the sensory and intellectual equipment used in observation. Epistemology thus presents physics with a new method of achieving its aims.  The development of relativity theory, and the transformation of quantum theory from an empirical to a rational theory are the outcome of the new method;  and in it is our great hope of further fundamental advances.  
------

Eddington was a bit overly idealistic in his description of scientists and the onlookers they so often reject a lot more rudely than that.  It's been my experience that it is a very rare scientist who has such a clear understanding of science and its inescapable limits, not to mention the humility to accept that their product is subjective and limited.  Most of the scientists I've read or encountered maintain an unbounded and all encompassing faith in the objective truth of what is presented as science by those people and institutions that they approve of, and their rejection of the idea that anything else could possibly "exist".  "Existence" in general use seems to mean "that which it is respectable to believe and which will not invoke our scorn".   Later in the book Eddington says some interesting things about the vaugery of the idea of "existence", maintaining that it is an obsession of philosophy but which doesn't have any real place or purpose in science.  Try telling that to the science blogging community.

Eddington was an idealist.  He had enormously important things to say about science and scientists but his ideal, while it might have applied to some, hardly applies to many of even the most august physical scientists today.  The culture of science has not kept up with the discoveries of physics in his generation.  I would point out that it certainly doesn't apply to those alleged sciences that you might expect would be most closely allied to epistemology.  Perhaps I'll get around to relating the story of the dead salmon whose brain activity showed up on an fMRI picture.