“Theory” is not an ideal choice of word for when certainty is meant to be implied. To be “just” a theory implies that a conclusion is untested, fraught with unseen holes, or suitable for debate only among wizened, shriveled men with mismatched socks and mild dementia. One may recall with amusement how Humboldt's contemporaries theorized on the chemical origins of life, seized by the notion that one element or another could alone twitch the limbs and animate the body (Dettelbach 13). Humboldt's rejection of the theory-centered approach seems in the context not only fortunate but prudent. Enthusiasm for this conclusion, however, may show signs of fizzling, however, the more one notices the similarities between that approach and the scientific method used today. Theory is far from unworkable and, in the context of modern science, may in practical terms be the only model that can be called feasible.
Humboldt's method of observation and comparison is lent an air of rigor for its apparent thoroughness. It certainly seems sound; once Humboldt's train of thought is understood, his detractors, their credibility slipping out from underneath them, look as though they just don't get it. Testing a large number of possible cases, in addition to generating many potential leads, has definite advantages over drawing generalizations from incomplete data.
But is that not the modus operandi of modern science? Humboldt lived prior to the advent of the compartmentalized, professionalized discipline today recognized as science, and with the increase in scope that accompanied that shift came problems of resources. Limited manpower, finite supplies, limited subject pools, and far from inexhaustible sources of funding altered the rules of the game. With the ability to test vast experimental groups came the mulling over of how small and how not necessarily representative a sample could be made while keeping conclusions legitimate and broadly applicable. Science became an enterprise, and there arose the question of how to run it efficiently.
The pursuit of objective truth is, by nature, an exercise in approximation. A psychological study isn't conducted just to figure out how a small group of specific individuals will behave under certain situations; the results of that study are applied to a population thousands of times the size of the group tested. Mice aren't poked and prodded with needles out of scientists' morbid curiosity about the results of deadly pathogens on rodents; rather, the mice serve as approximations of humans, and the results of their infection serve as approximations for the effects of disease on human patients. Naturalistic observation, the method of research closest to that employed by Humboldt, yields data that is taken with small dunes of salt because there of the absence of repeatability and the impossibility of identifying an innumerable horde of confounding variables.
To carry out experiments following Alexander von Humboldt's example in the age of modern science would therefore not only be time-consuming, but wasteful. Liberated for the most part from its faults by institutions, collaboration, and peer review, theory-minded research has, for its utility as much as its results, established itself as the operational paradigm of this day and age.
Dettelbach, Michael. “Alexander von Humboldt between Enlightenment and Romanticism.”
Northeastern Naturalist, Vol. 8, Special Issue 1: Alexander von Humboldt's Natural History Legacy and Its Relevance for Today (2001): pp. 9-20. JSTOR. Web. 16 January 2011