You are on page 1of 33

The Function of Measurement in Modern Physical Science

B y Thomas S. K u h n * T the U niversity of Chicago, the facade of the Social Science Research Building bears Lord K elvins famous dictum : If you cannot measure, your knowledge is meager and unsatisfactory. 1 W ould that statem ent be there if it had been written, not by a physicist, but by a sociologist, political scientist, or economist? Or again, would terms like meter reading and yardstick recur so frequently in contem porary discussions of epistemology and scientific method were it not for the prestige of modern physical science and the fact that measurement so obviously bulks large in its research? Suspecting that the answer to both these questions is no, I find my assigned role in this con ference particularly challenging. Because physical science is so often seen as the paradigm of sound knowledge and because quantitative techniques seem to provide an essential clue to its success, the question how measurement hits actually functioned for the past three centuries in physical science arouses more than its natural and intrinsic interest. Let me therefore make my gen eral position clear at the start. Both as an ex-physicist and as an historian of physical science I feel sure that, for a t least a century and a half, quantitative methods have indeed been central to the development of the fields 1 study. On the other hand. I feel equally convinccd th a t our most prevalent notions both about the function of measurement and about the source of its special efficacy arc derived largely from m yth. P artly because of this conviction and partly for more autobiographical rea sons,2 I shall employ in this paper an approach rather different from th at of most other contributors to this conference. U ntil almost its close my essay will include no narrative of the increasing deployment of q uantitative tech niques in physical science since the close of the M iddle Ages. Instead, the two
U n iv ersity of C alifornia, Berkeley 1 F o r the facade see, E leven T w e n ty - S ix : A Decaile o f Social Science R esearch, ed. Louis W irth (C hicago, 1940), p. 169. T h e sentim ent there inscribed recurs in K elvins w ritings, but I have found no form ulation eloser to the Chicago quotation than the follow ing: W h en you cannot express it in num bers, your know ledge is of a m eagre an d unsatisfactory kind. See S ir W illiam Tljom son, E lectrical U n its of M easurem ent," P opular L ectures and A ddresses, 3 vols. (London, 18899 1 ), I, 73. 2 T h e central sections of this paper, which w as added to the present p ro g ram a t a late date, a re abstracted from m y essay, T he Role of M easurem ent in tfie D evelopm ent of N a tu ral Science, a m ultilithed revision o f a talk first given to the Social Sciences Colloquium of the U n iv ersity of C alifornia, B erkeley. T h at version w ill be published in a volum e of papers on Q uantification in th e Social Sciences th at g ro w s o u t of th e Berkeley colloquium . In deriv in g the present pap er from it, I have p repared a new introduction and last section, and have som ew hat condensed the m aterial th a t intervenes.




central questions of this paper how has m easurem ent actually functioned in physical science, and w hat has been the source of its special efficacy will be approached directly. For this purpose, and for it alone, history will truly be philosophy teaching by example. Before perm itting history to function even as a source of examples, we must, however, grasp the full significance of allowing it any function a t all. To th at end my paper opens with a critical discussion of w hat I take to be the most prevalent image of scientific m easurem ent, an image th at gains much of its plausibility and force from the m anner in which com putation and meas urem ent enter into a profoundly unhistorical source, the science text. T h at discussion, confined to Section I below, will suggest th at there is a textbook image or m yth of science and th at it m ay be system atically misleading. M eas urem ents actual function either in the search for new theories or in the confirmation of those already at hand m ust be sought in the journal litera ture, which displays not finished and accepted theories, but theories in the process of development. A fter th at point in the discussion, history will nec essarily become our guide, and Sections I I and I I I will attem p t to present a more valid image of m easurem ents most usual functions drawn from th at source. Section IV employs the resulting description to ask why measurem ent should have proved so extraordinarily effective in physical research. Only after th at, in the concluding section, shall I attem p t a synoptic view of the route by which measurement has come increasingly to dom inate physical sci ence during the past three hundred years. [One more caveat proves necessary before beginning. A few participants in this conference seem occasionally to mean by measurem ent any unam bigu ous scientific experiment or observation. T hus, Professor Boring supposes that D escartes was measuring when he dem onstrated the inverted retinal image a t the back of the eye-ball; presum ably he would say the same about F ranklins dem onstration of the opposite polarity of the two coatings on a Leyden jar. Now I have no doubt th at experim ents like these are among the most significant and fundam ental th a t the physical sciences have known, but I see no virtue in describing their results as measurements. In any case, that terminology would obscure what are perhaps the most im portant points to be made in this paper. I shall therefore suppose th at a measurement (or a fully quantified theory) always produces actual numbers. Experim ents like D escartes or F ranklins, above, will be classified as qualitative or as nonnumerical, w ithout, I hope, a t all implying th a t they are therefore less im portant. Only with th a t distinction between qualitative and q u antitative avail able can I hope to show th a t large am ounts of qualitative work have usually been prerequisite to fruitful quantification in the physical sciences. And only if th at point can be made shall we be in a position even to ask about the effects of introducing quantitative methods into sciences th at had previously proceeded w ithout m ajor assistance from them .] I. T E X T B O O K M E A SU R E M E N T T o a very much greater extent than we ordinarily realize, our image of physi cal science and of m easurem ent is conditioned by science texts. In p a rt th at



influence is direct: textbooks are the sole source of most peoples firsthand acquaintance with the physical sciences. T heir indirect iniluence is, however, undoubtedly larger and more pervasive. Textbooks or their equivalent are the unique repository of the finished achievements of modern physical scien tists. It is with the analysis and propagation of these achievements th at most writings on the philosophy of science and most interpretations of science for the nonscientist arc concerned. As many autobiographies attest, even the research scientist does not always free himself from the textbook image gained during his first exposures to science.8 I shall shortly indicate why the textbook mode of presentation must in evitably be misleading, but let us first examine th a t presentation itself. Since most participants in this conference have already been exposed to at least one textbook of physical science, I restrict attention to the schematic trip artite sum m ary in the following figure. I t displays, in the upper left, a series of Theory (x ) 0i (x) (x) 02(x )

M anipulation (Logic and M ath )

theoretical and lawlike statem ents, (x) 0 i(x ), which together constitute the theory of the science being described.4 T he center of the diagram represents the logical and mathematical equipment employed in m anipulating the theory. Law like statem ents from the upper left are to be imagined fed into the
3 T h is phenom enon is exam ined in m ore detail in m y m onograph, T h e S tru c tu re of S cientific R evolutions, to appear w hen com pleted a s Vol. IT. No. 2, in the International Encyclopedia o f U nified Science. M any other aspects of the textbook im age of science, its sourccs and its strengths, a r c also exam ined in th a t place. 4 O bviously n o t all the statem ents required to constitute m ost theories a rc of th is p articu la r logical form , but the com plexities have no relevance to the points m ade here. R . B. B raithw aite, S cien tific E xp la n a tio n (C a m bridge, E ngland, 1953) includes a useful, though very general, description of th e logi cal stru c tu re of scientific theories.



hopper a t the top of the machine together with certain initial conditions specifying the situation to which the theory is being applied. T h e crank is then turned; logical and mathematical operations are internally perform ed; and numerical predictions for the application at hand emerge in the chute at the front of the machine. These predictions are entered in the left-hand col umn of the table th at appears in the lower right of the figure. T h e right-hand column contains the numerical results of actual m easurem ents, placed there so th a t they may be compared with the predictions derived from the theory. M ost texts of physics, chem istry, astronom y, etc. contain many d ata of this sort, though they are not always presented in tabular form. Some of you will, for example, be more fam iliar with equivalent graphical presentations. T he table a t the lower right is of particular concern, for it is there that the results of measurement appear explicitly. W hat may we take to be the sig nificance of such a table and of the num bers it contains? I suppose th at there are two usual answ ers: the first, immediate and almost universal; the other, perhaps more im portant, but very rarely explicit. M ost obviously the results in the table seem to function as a test of theory. If corresponding num bers in the two columns agree, the theory is acceptable; if they do not, the theory must be modified or rejected. T his is the function of measurement as confirmation, here seen emerging, as it does for most read ers, from the textbook formulation of a finished scientific theory. For the time being I shall assum e that some such function is also regularly exempli fied in norm al scientific practice and can be isolated in w ritings whose p u r pose is not exclusively pedagogic. At this point we need only notice th at on the question of practice, textbooks provide no evidence whatsoever. N o text book ever included a table th a t either intended or managed to infirm the the ory the text was w ritten to describe. R eaders of current science texts accept the theories there expounded on the authority of the author and the scientific community, not because of any tables th a t these texts contain. If the tables are read at all, as they often are, they are read for another reason. I shall inquire for this other reason in a moment b u t must first rem ark on the second putative function of measurem ent, th a t of exploration. N um erical d ata like those collected in the right-hand column of our table can, it is often supposed, be useful in suggesting new scientific theories or laws. Some people seem to take for granted th a t numerical d ata are more likely to be productive of new generalizations than any other sort. It is th at special p ro ductivity, rather than m easurem ents function in confirmation, th at probably accounts for K elvins dictum s being inscribed on the facade a t the U niversity of Chicago.6 I t is by no means obvious th a t our ideas about this function of numbers are related to the textbook schema outlined in the diagram above, yet I see no other way to account for the special efficacy often attributed to the results of measurement. W e are, I suspect, here confronted with a vestige of an ad m ittedly outworn belief that laws and theories can be arrived a t by some
8 P ro fesso r F ra n k K night, for exam ple, suggcsts th a t to social scientists the practical m eaning [of K elvins statem ent) tends to be: I f you cannot m easure, m easure anyhow . E leven T w e n ty -S ix , p. 169.



process like running the machine backw ards. Given the numerical d ata in the Experim ent column of the table, logico-mathematical manipulation (aided, all would now insist, by intuition ) can proceed to the statem ent of the laws th at underlie the numbers. If any process even rem otely like this is involved in discovery if, th a t is, laws and theories are forged directly from d ata by the mind then the superiority of numerical to qualitative data is immediately apparent. T h e results of measurement are neutral and precise; they cannot mislead. Even more im portant, numbers are subject to m athe matical m anipulation; more than any other form of data, they can be assimi lated to the semimechanical textbook schema. I have already implied my skepticism about these two prevalent descrip tions of the function of measurement. In Sections I I and I I I each of these functions will be further compared w ith ordinary scientific practice. B ut it will help first critically to pursue our examination of textbook tables. By doing so I would hope to suggest th at our stereotypes about measurement do not even quite fit the textbook schema from which they seem to derive. Though the num erical tables in a textbook do not there function either for exploration or confirmation, they are there for a reason. T h a t reason we may perhaps discover by asking w hat the author of a text can mean when he says th at the num bers in the T heory and Experim ent column of a table agree. At best the criterion must be in agreement within the limits of accuracy of the m easuring instrum ents employed. Since com putation from theory can usually be pushed to any desired num ber of decimal places, exact or num eri cal agreement is impossible in principle. B ut anyone who has examined the tables in which the results of theory and experiment are compared must rec ognize th a t agreem ent of this more modest sort is rather rare. Almost always the application of a physical theory involves some approxim ation (in fact, the plane is not frictionless, the vacuum is not perfect, the atom s are not unaffected by collisions), and the theory is not therefore expected to yield quite precise results. Or the construction of the instrum ent may involve ap proxim ations (e.g., the linearity of vacuum tube characteristics) th at cast doubt upon the significance of the last decimal place th at can be unam bigu ously read from their dial. Or it may simply be recognized th at, for reasons not clearly understood, the theory whose results have been tabulated or the instrum ent used in m easurem ent provides only estimates. For one of these reasons or another, physical scientists rarely expect agreem ent quite within instrum ental limits. In fact, they often distrust it when they see it. A t least on a student lab report overly close agreem ent is usually taken as presum p tive evidence of d ata m anipulation. T h a t no experiment gives quite the ex pected numerical result is sometimes called T he F ifth Law of Therm ody namics.T he fact that, unlike some other scientific laws, it has acknowledged exceptions does not diminish its utility as a guiding principle. It follows th a t w hat scientists seek in numerical tables is not usually agreeT h e first three L aw s of T herm odynam ics a rc well know n outside the trade. T h e F o u rth L aw " states th a t no piece of experim ental ap p aratu s w o rk s th e first tim e it is set up. W e shall exam ine evidence fo r the F ifth L aw below.



m ent at all, but w hat they often call reasonable agreem ent. Furtherm ore, if we now ask for a criterion of reasonable agreem ent, we are literally forced to look in the tables themselves. Scientific practice exhibits no consistently applied or consistently applicable external criterion. Reasonable agreem ent varies from one p a rt of science to another, and within any p a rt of science it varies with time. W hat to Ptolem y and his immediate successors was reason able agreement between astronom ical theory and observation was to Coper nicus incisive evidence th at the Ptolem aic system m ust be wrong.7 Between the times of Cavendish (1731-1810) and Ram say (1852-1916), a similar change in accepted chemical criteria for reasonable agreem ent led to the study of the noble gases.8 T hese divergences are typical and they are matched by those between contem porary branches of the scientific community. In p arts of spectroscopy reasonable agreem ent means agreement in the first six or eight left-hand digits in the num bers of a table of wave lengths. In the theory of solids, by contrast, two-place agreem ent is often considered very good in deed. Yet there are p arts of astronom y in which any search for even so lim ited an agreem ent m ust seem utopian. In the theoretical study of stellar mag nitudes agreement to a m ultiplicative factor of ten is often taken to be reasonable. N otice th at we have now inadvertently answered the question from which we began. We have, th a t is, said what agreem ent between theory and ex perim ent m ust mean if th a t criterion is to be draw n from the tables of a science text. B ut in doing so we have gone full circle. I began by asking, at least by implication, what characteristic the num bers of the table must ex hibit if they are to be said to agree. I now conclude th at the only possible criterion is the mere fact that they appear, together with the theory from which they arc derived, in a professionally accepted text. W hen they appear in a text, tables of num bers drawn from theory and experim ents cannot dem onstrate anything but reasonable agreem ent. And even th at they demon strate only by tautology, since they alone provide the definition of reasonable agreem ent that has been accepted by the profession. T h a t, I think, is why the tables are there: they define reasonable agreem ent. B y studying them, the reader learns w hat can be expected of the theory. An acquaintance with the tables is p art of an acquaintance with the theory itself. W ithout the tables, the theory would be essentially incomplete. W ith respect to measure m ent, it would be not so much untested as untestable. W hich brings us very close to the conclusion th a t, once it has been embodied in a text which for present purposes means, once it has been adopted by the professionno th e ory is recognized to be testable by any quantitative tests th at it has n o t al ready passed.9
7 T . S. K uhn, T h e C opem ican R evolution (C am bridge, M ass., 1957), pp. 72-76, 135-143. 8 W illiam R am say, T h e G ases o f the A t tnosphcre: the H isto r y o f T h e ir D iscovery (L ondon, 1896), C hapters 4 and 5. * T o pursue th is point w ould c a rry us far beyond the subject of th is paper, but it should bc pursued because, if I am rig h t, it relates to th e im portant contem porary controversy over the distinction betw een an aly tic and synthetic tru th . T o th e ex ten t th a t a scientific theory m ust be accom panied by a statem en t of the evidence fo r it in o rd e r to have em pirical m eaning, the full th eo ry (w h ich includes the evidence) m ust be analytically tru e. For a statem ent of the philosophical problem of an alyticity sec W . V. Q uine, T w o D ogm as of E m piricism and o th er essays in F ro m a L o g i -



Perhaps these conclusions are not surprising. Certainly they should not be. Textbooks are, after all, w ritten some time after the discoveries and confirma tion procedures whose outcomes they record. Furtherm ore, they are written for purposes of pedagogy. T he objective of a textbook is to provide the reader, in the most economical and easily assimilable form, w ith a statem ent of w hat the contem porary scientific community believes it knows and of the principal uses to which th at knowledge can be put. Inform ation about the ways in which th at knowledge was acquired (discovery) and in which it was enforced on the profession (confirm ation) would at best be excess baggage. Though including th a t inform ation would almost certainly increase the hum anistic values of the text and might conceivably breed more flexible and creative scientists, it would inevitably detract from the ease of learning the contem porary scien tific language. T o date only the last objective has been taken seriously by most w riters of textbooks in the natural sciences. As a result, though texts m ay be the right place for philosophers to discover the logical structure of finished scientific theories, they are more likely to mislead than to help the unw ary individual who asks about productive methods. One might equally appropriately go to a college language text for an authoritative characteriza tion of the corresponding literature. Language texts, like science texts, teach how to read literature, not how to create or evaluate it. W hat signposts they supply to these latter points are most likely to point in the wrong direction.10 II. M O T IV ES FO R N O R M A L M E A S U R E M E N T T hese considerations dictate our next step. W e m ust ask how measurement comes to be juxtaposed with laws and theories in science texts. Furtherm ore, we must go for an answer to the journal literature, the medium through which natural scientists report their own original work and in which they evaluate th at done by others. Recourse to this body of literature immediately casts doubt upon one implication of the standard textbook schema. Only a minis cule fraction of even the best and most creative m easurements undertaken by
cal P o in t o f V iezv (C am bridge, M ass., 1953). F o r a stim ulating, but loose, discussion of the occasionally analytic sta tu s of sdcntific law s, see N . R . H anson, P a tte rn s o f D iscovery (C am bridge, E ngland, 1958), pp. 93-118. A new discussion of th e philosophical problem , including copious references to th e controversial literature, is A lan Pasch, E xperience and the A n a ly tic : A R econsideration o f E m p irictsnt (C hicago, 1958). 10 T h e m onograph cited in note 3 w ill a rg u e th a t the m isdirection supplied by science te x ts is both system atic and functional. It is by no m cans clear that a m ore accurate im age of the scientific processes would enhance the research efliciency of physical scientists. 11 It is, of course, som ew hat anachronistic to apply the term s jo u rn al lite ra tu re ' and textbooks" in the w hole of the period I have been asked to discuss. B ut I am concerned to em phasize a p attern of professional com m unication w hose o rig in s a t least can be found in the seventeenth ce n tu ry and w hich has increased in rig o r ever since. T h e re w as a tim e (differen t in different sciences) w hen the patte rn of com m unication in a science w as much the sam e a s th a t still visible in the hum anities and m any of the social sciences, but in all the physical sciences th is p attern is a t least a centu ry gone, and in m any of them it disappeared even e a rlie r than t h a t N ow all publication of research resu lts occurs in jo u rn als read only by the profession. Rooks are exclusively tc x tbooks, com pendia, popularizations, o r plulosophical reflections, and w ritin g them is a som ew hat suspect, because nonprofessional, activity. N eedless to say th is sh arp and rigid s e r r a t i o n betw een articles and books, research and nonrcsearch w ritings, g rea tly increases the stren g th of w hat I have callcd the te x tbook image.



natural scientists are m otivated by a desire to discover new q uantitative regu larities or to confirm old ones. Almost as small a fraction turn out to have had either of these effects. T here are a few th at did so, and I shall have something to say about them in Sections I I I and IV. But it will help first to discover just why these exploratory and confirmatory m easurements are so rare. In this section and most of the next, I therefore restrict myself to meas urem ents most usual function in the normal practice of science.12 Probably the rarest and most profound sort of genius in physical science is th at displayed by men who, like Newton, Lavoisier, or Einstein, enunciate a whole new theory th a t brings potential order to a vast num ber of natural phenomena. Yet radical reform ulations of this sort are extremely rare, largely because the state of science very seldom provides occasion for them. M ore over, they are not the only truly essential and creative events in the develop ment of scientific knowledge. T he new order provided by a revolutionary new theory in the natural sciences is always overwhelmingly a potential order. M uch work and skill, together with occasional genius, are required to make it actual. And actual it must be made, for only through the process of actu alization can occasions for new theoretical reform ulations be discovered. The bulk of scientific practice is thus a complex and consuming mopping-up op eration th a t consolidates the ground made available by the most recent theo retical breakthrough and th a t provides essential preparation for the break through to follow. In such mopping-up operations, m easurem ent has its overwhelmingly most common scientific function. Ju st how im portant and difficult these consolidating operations can be is indicated by the present state of E insteins general theory of relativity. The equations embodying th at theory have proved so difficult to apply th a t (ex cluding the limiting case in which the equations reduce to those of special relativity) they have so far yielded only three predictions th a t can be com pared with observation.18 M en of undoubted genius have totally failed to develop others, and the problem rem ains worth their attention. Until it is solved, E insteins general theory rem ains a largely fruitless, because unexploitable, achievem ent.1 4 U ndoubtedly the general theory of relativity is an extrem e case, but the situation it illustrates is typical. Consider, for a somewhat more extended example, the problem th a t engaged much of the best eighteenth-century scien
12 H e re and elsew here in th is paper I ignore th e v ery la rg e am ount of m easurem ent done sim ply to g ath er factual inform ation. I think of such m easurem ents a s specific gravities, w ave lengths, sprin g constants, boiling points, ctc undertaken in o rd e r to determ ine param ete rs that m ust be inserted into scientific Uieories but w hose num erical outcom e those thcories do not ( o r did not in the relevant p eriod ) predict. T h is sort of m easurem ent is not w ith out interest, hut I thin k it w idely understood, In an y case, considering it w ould too g rea tly extend the lim its of this paper. 13T hese a r e : th e deflection of light in the suns gravitational field, the procession of the perihelion of M ercu ry , and th e red shift of light from d istan t sta rs. O nly th e first tw o a re actu ally q u an titativ e predictions in the present sta te of the theory, 14 T h e difficulties in producing concrete applications of th e general theory of relativ ity need n o t prevent scientists from attem p tin g to exploit the scicntific view point embodied in th a t theory. But, p erh ap s u nfortunately, it seems to be doing so. U n lik e th e special thco ry, general relativ ity is today very little studied by stu d en ts of physics. W ith in fifty years we m ay conceivably have to tally lost sig h t of th is aspect of E in ste in 's contribution,



tific thought, th a t of deriving testable numerical predictions from N ew tons three Laws of motion and from his principle of universal gravitation. W hen N ew tons theory was first enunciated late in the seventeenth century, only his T hird Law (equality of action and reaction) could be directly investigated by experiment, and the relevant experim ents applied only to very special cases.15 T he first direct and unequivocal dem onstrations of the Second Law awaited the development of the Atwood machine, a subtly conceived piece of labora tory apparatus th at was not invented until almost a century after the appear ance of the Principia .* D irect quantitative investigations of gravitational attraction proved even more difficult and were not presented in the scientific literature until 1798.11 N ew tons F irst Law cannot, to this day, be directly compared with the results of laboratory measurement, though developments in rocketry make it likely that we have not much longer to wait. I t is, of course, direct dem onstrations, like those of Atwood, th at figure most largely in natural science texts and in elem entary laboratory exercises. Be cause simple and unequivocal, they have the greatest pedagogic value. T h at they were not and could scarcely have been available for more than a century a fter the publication of N ew tons work m akes no pedagogic difference. A t most it only leads us to m istake the nature of scientific achievement.1* * B ut if N ew tons contem poraries and successors had been forced to wait th a t long for quantitative evidence, apparatus capable of providing it would never have been designed. Fortunately there was another route, and much eigthteenthcentury scientific talent followed it. Complex m athem atical manipulations, exploiting all the laws together, perm itted a few other sorts of prediction cap able of being compared with quantitative observation, particularly with lab oratory observations of pendula and with astronomical observations of the motions of the moon and planets. B ut these predictions presented another and equally severe problem, th at of essential approxim ations.1* T he suspen15 T h e m ost relevant and widely em ployed experim ents w ere perform ed w ith pendula. D eterm ination of the recoil w hen tw o pendulum bobs collided seems to have been the m ain conceptual and experim ental tool used in the seventeenth century to determ ine w hat dynam ical "a ctio n ' and "reaction w ere. See A . W olf, A H isto ry o f Science, T echnology, and P hilosophy in the 16th & 17th C enturies, new cd. prepared by D . M cK ic (London, 1950), pp. 155, 231-235: and R. D ugas, L a m ic a n iq u t au xvii* siecle (N euchatel, 1954), pp. 283-298; and S ir Isaac N n v to n 's M athetnatical P rincipies o f N a tu ra l P hilosophy and h is S y s te m o f th e W orld, cd. F . C a jo ri (B erkeley, 1934), pp. 21-28. W olf (p. 155) describes the T h ird L aw a s "th e only physical law of the three. 16 See the excellent description of th is app aratus and the discussion of A tw oods reasons for building it in H anson, P a ttern s o f D iscoi'ery, pp. 100-102 ami notes to these pages. 17 A . W olf, A H isto ry o f Science, T echnology, and P hilosophy in the E ighteenth C entu ry, 2nd cd. revised by D . M cK ic (L ondon, 1952), pp. 111-113. T h e re a r e some p recurso rs o f C avendishs m easurem ents o f 1798, b u t it is only a fte r Cavendish th at m easurem ent begins to yield consistent results. ** M odem lab o rato ry ap p a ratu s designed to help students study G alileos law of free fall provides a classic, though perhaps quite necessary, exam ple of th e w ay pedagogy m isdirects the historical im agination about the relation bctw een creativ e science and m easurem ent. None of the ap p a ratu s now used could possibly have been built in the seventeenth century. O ne of the best and m ost widely dissem inated pieces of equipm ent, fo r exam ple, allow s a heavy bob to fall between a p air of parallel vertical rails. T h ese rails a r c electrically charg ed every l/1 0 0 th of a second, and th e sp ark th a t then passes th ro u g h th e bob from rail to rail records th e bobs position on a chem ically tre a tcd tape. O th e r picccs of ap p a ratu s involve electric tim ers, etc. F o r the h isto rical difiiculties of m aking m easurem ents relevant to this law , see below. 19 A ll th e applications of N ew to n s Law s involve appro x im atio n s of som e so rt, but in the follow ing exam ples the app ro x im atio n s



sions of laboratory pendula are neither weightless nor perfectly elastic; air resistance dam ps the motion of the bob; besides, the bob itself is of finite size, and there is the question of which point of the bob should be used in com put ing the pendulum s length. If these three aspects of the experim ental situation are neglected, only the roughest sort of quantitative agreem ent between theory and observation can be expected. B ut determining how to reduce them (only the last is fully eliminable) and w hat allowance to make for the residue are themselves problems of the utm ost difficulty. Since N ew tons day much bril liant research has been devoted to their challenge.20 The problems encountered when applying N ew tons Laws to astronomical prediction are even more revealing. Since each of the bodies in the solar sys tem a ttracts and is attracted by every other, precise prediction of celestial phenom ena demanded, in N ew tons day, the application of his Laws to the simultaneous motions and interactions of eight celestial bodies. (T hese were the sun, moon, and six known planets. I ignore the other planetary satellites.) T he result is a m athem atical problem th at has never been solved exactly. To get equations th a t could be solved, Newton was forced to the simplifying a s sumption th at each of the planets was attracted only by the sun, and the moon only by the earth. W ith this assumption, he was able to derive K eplers famous Laws, a wonderfully convincing argum ent for his theory. But deviation of planets from the motions predicted by K eplers Laws is quite apparent to simple quantitative telescopic observation. T o discover how to treat these deviations by Newtonian theory, it was necessary to devise mathematical estim ates of the perturbations produced in a basically Keplerian o rb it by the interplanetary forces neglected in the initial derivation of K eplers Laws. N ew tons mathematical genius was displayed at its best when he produced a first crude estim ate for the perturbation of the moons motion caused by the sun. Im proving his answer and developing similar approxim ate answers for the planets exercised the greatest m athem atical minds of the eighteenth and early nineteenth centuries, including those of Euler, Lagrange, Laplace, and Gauss.21 Only as a result of their work was it possible to recognize the anom aly in M ercurys motion th a t was ultim ately to be explained by E insteins general theory. T h at anomaly had previously been hidden within th e limits of reasonable agreem ent. As far as it goes, the situation illustrated by quantitative application of N ew tons Laws is, I think perfectly typical. Similar examples could be p ro duced from the history of the corpuscular, the wave, or the quantum mechan ical theory of light, from the history of electrom agnetic theory, quantitative chemical analysis, or any other of the numerous natural scientific theories with quantitative implications. In each of these cases, it proved immensely difficult to find many problems th at perm itted quantitative comparison of theory and observation. Even when such problems were found, the highest scientific tal ents were often required to invent apparatus, reduce perturbing effects, and
liave a quantitative im portance that they do n ot possess in those th a t precede. 20 W olf, E ighteenth C entury, pp. 75-81, provides a good prelim inary description of this w ork. 21 Ibid., pp. 96-101. W illiam W h ew ell, H istn ry o f th e In d u ctive Sciences, rev. ed., 3 vols. (L ondon, 1847), II, 213-271.



estim ate the allowance to be made for those th a t remained. This is the sort of work th at most physical scientists do most of the time insofar as their work is quantitative. Its objective is, on the one hand, to improve th e measure of reasonable agreem ent characteristic of the theory in a given application and, on the other, to open up new areas of application and establish new measures of reasonable agreem ent applicable to them. For anyone who finds m athe matical or manipulative puzzles challenging, this can be fascinating and in tensely rewarding work. And there is always the remote possibility th a t it will pay an additional dividend: something may go wrong. Y et unless something does go wrong a situation to be explored in Section IV these finer and finer investigations of the q uantitative m atch between theory and observation cannot be described as attem pts a t discovery or a t confirmation. T h e man who is successful proves his talents, but he does so by getting a result th a t the entire scientific community had anticipated some one would someday achieve. H is success lies only in th e explicit dem onstra tion of a previously implicit agreem ent between theory and the world. N o novelty in n ature has been revealed. N or can the scientist who is successful in this sort of work quite be said to have confirmed the theory th a t guided his research. F o r if success in his venture confirms the theory, then failure ought certainly infirm it, and nothing of the sort is tru e in this case. F ail ure to solve one of these puzzles counts only against the scientist; he has p u t in a great deal of time on a project whose outcome is not worth publication; the conclusion to be drawn, if any, is only th a t his talents were not adequate to it. I f m easurem ent ever leads to discovery or to confirmation, it does not do so in the most usual of all its applications. I I I . T H E E F F E C T S O F N O R M A L M E A SU R E M E N T T here is a second significant aspect of the norm al problem of measurement in natural science. So far we have considered why scientists usually measure; nowr we m ust consider the results th a t they get when they do so. Im m ed iately another stereotype enforced by textbooks is called in question. In textbooks the numbers th a t result from measurem ent usually appear as the archetypes of the irreducible and stubborn facts to which the scientist must, by struggle, m ake his theories conform. B ut in scientific practice, as seen through the journal literature, the scientist often seems rather to be strug gling with facts, trying to force them into conform ity with a theory he does not doubt. Q uantitative facts cease to seem simply the given. T hey m ust be fought for and with, and in this fight the theory with which they are to be compared proves the most potent weapon. O ften scientists cannot get numbers th at compare well with theory until they know w hat num bers they should be making nature yield. P a rt of this problem is simply the difficulty in finding techniques and instru ments th at perm it the comparison of theory with quantitative measurements. We have already seen th a t it took almost a century to invent a machine th at could give a straightforw ard quantitative dem onstration of Newtons Second Law. Hut the machine th at Charles Atwood described in 1784 was not the first instrum ent to yield quantitative inform ation relevant to th at Law.



A ttem pts in this direction had been made ever since Galileos description of his classic inclined plane experiment in 1638.28 Galileos brilliant intuition had seen in this laboratory device a way of investigating how a body moves when acted upon only by its own weight. A fter the experim ent he announced th at measurement of the distance covered in a m easured time by a sphere rolling down the plane confirmed his prior thesis th a t the motion was uniformly accelerated. As reinterpreted by Newton, this result exemplified the Second Law for the special case of a uniform force. B ut Galileo did not report the num bers he had gotten, and a group of the best scientists in F rance announced their total failure to get com parable results. In p rin t they wondered w hether Galileo could himself have tried the experiment. In fact, it is almost certain th a t Galileo did perform th e experiment. If he did, he must surely have gotten quantitative results th a t seemed to him in adequate agreem ent w ith the law ( s = V & at2) th a t he had shown to be a con sequence of uniform acceleration. But anyone who has noted the stop-watches or electric timers, and the long planes or heavy llywheels needed to perform this experim ent in modern elem entary laboratories may legitimately suspect th a t Galileos results were not in unequivocal agreem ent w ith his law. Quite possibly the French group looking even a t the same d ata would have wondered how they could seem to exemplify uniform acceleration. This is, of course, largely speculation. B ut the speculative elem ent casts no doubt upon my present point: whatever its source, disagreem ent between Galileo and those who tried to repeat his experiment was entirely natural. If Galileos generali zation had not sent men to the very border of existing instrum entation, an area in which experimental scatter and disagreement about interpretation were inevitable, then no genius would have been required to m ake it. His example typifies one im portant aspect of theoretical genius in the natural sciences it is a genius th at leaps ahead of the facts, leaving the rather different talent of the experim entalist and instrum entalist to catch up. In this case catching up took a long time. T he Atwood M achine was designed because, in the middle of the eighteenth century, some of the best Continental scientists still wondered whether acceleration provided the proper m easure of force. Though their doubts derived from more than measurement, m easurement was still sufficiently equivocal to fit a variety of different quantitative conclusions.** The preceding example illustrates the difficulties and displays the role of theory in reducing scatter in the results of measurement. T here is, however, more to the problem. W hen measurement is insecure, one of the tests for reliability of existing instrum ents and m anipulative techniques m ust inevitably be their ability to give results th at compare favorably w ith existing theory. In some p arts of natural science, the adequacy of experimental technique can be judged only in this way. W hen th a t occurs, one may not even speak of in secure instrum entation or technique, implying th a t these could be improved without recourse to an external theoretical standard.
22 F o r a m odern E nglish version of the o riginal sec G alileo G alilei, D ialogues C oncernm g T w o N e w Sciences, tran s. H en ry C rew and A . D e S alvio (E v a n sto n and Chicago, 1946), pp. 171-172. 23 T h is w hole sto ry an d m ore is b rillian tly set fo rth in A . K o y ri, "A n E x p erim en t in M easurem ent, P roc. A m c r. P hil. S o c., 1953, 97: 222-237. 21 H anson, P a tte rn s o f D iscovery, p. 101.



For example, when John D alton first conceived of using chemical m easure m ents to elaborate an atom ic theory th a t he had initially drawn from meteoro logical and physical observations, he began by searching th e existing chemical literature for relevant data. Soon he realized th at significant illumination could be obtained from those groups of reactions in which a single pair of elements, e.g., nitrogen and oxygen, entered into more than one chemical com bination. If his atomic theory were right, the constituent molecules of these compounds should differ only in the ratio of the num ber of whole atom s of each element th a t they contained. T he th ree oxides of nitrogen m ight, for example, have molecules N 20 , NO, and NO*, or they might have some other similarly simple arrangem ent.25 B ut whatever the particular arrangem ents, if the weight of nitrogen were the same in th e samples of the three oxides, then the weights of oxygen in the three samples should be related to each other by simple whole-number proportions. Generalization of this principle to all groups of compounds formed from the same group of elem ents produced D altons Law of M ultiple Proportions. Needless to say, D altons search of the literature yielded some d ata th at, in his view, sufficiently supported the Law. But and this is the point of the illustration much of the then extant d ata did not support D altons Law a t all. F o r example, the measurem ents of the French chem ist P roust on the two oxides of copper yielded, for a given weight of copper, a weight ratio for oxygen of 1.47:1. On D altons theory the ratio ought to have been 2:1, and P ro u st is ju st the chemist who m ight have been expected to confirm the prediction. He was, in the first place, a fine experimentalist. Besides, he was then engaged in a m ajor controversy involving the oxides of copper, a controversy in which he upheld a view very close to D altons. B ut, at the beginning of the nineteenth century, chemists did not know how to perform q uantitative analyses that displayed multiple proportions. By 1850 they had learned, but only by letting D altons theory lead them. Knowing w hat results they should expect from chemical analyses, chemists were able to devise techniques th a t got them. As a result chem istry texts can now state th a t quantitative analysis confirms D altons atomism and forget that, historically, the relevant analytic techniques are based upon the very theory they are said to confirm. Before D altons theory was announced, measurement did not give the same results. T here are self-fulfilling prophecies in the physical as well as in the social sciences. T h a t example seems to me quite typical of the way measurement responds to theory in many p arts of the natural sciences. I am less sure th at my next, and far stranger, example is equally typical, but colleagues in nuclear physics assure me th at they repeatedly encounter sim ilar irreversible shifts in the results of measurement.
25 T h is is not, of course, D altons original notation. In fact, I am som ew hat m odernizing an d sim plifying th is w hole account. It can be reconstructed m ore fully fro m : A. N . M eldrum , T h e D evelopm ent of the A tom ic T h eo r y : (1 ) B erthollets D octrine of V aria b le P roportions, M anch. M em ., 1910, 54: 1-16; and (6 ) T h e R eception accorded to the T h eory advocated by D alton. ibid., 1911, 55: 1-10; L. K . N ash, T h e A to m ic M olecular Theory. H a rv a rd C ase H isto rie s in E x p erim en tal Science, C ase 4 (C am bridge, M ass., 1950") ; and T h e O rig in s of D altons Chem ical A tom ic T h eo ry , Isis, 1956, 47: 110-116. See also the useful discussions of atom ic w eight scattered th ro u g h J . R . P artin g to n , A S h o rt H isto r y of C h em istry, 2nd ed. (1-ondon, 1951).



Very early in the nineteenth century, P. S. de Laplace, perhaps the greatest and certainly the most famous physicist of his day, suggested th at the recently observed heating of a gas when rapidly compressed m ight explain one of the outstanding num erical discrepancies of theoretical physics. T his was the dis agreement, approxim ately 20 per cent, between the predicted and measured values of the speed of sound in air a discrepancy th at had attracted the a t tention of all E uropes best m athem atical physicists since Newton had first pointed it out. W hen Laplaces suggestion was made, it defied numerical con firmation (note the recurrence of this typical difficulty), because it demanded refined measurements of the therm al properties of gases, m easurem ents th at were beyond the capacity of apparatus designed for m easurem ents on solids and liquids. B ut the French Academy offered a prize for such m easurem ents, and in 1819 the prize was won by two brilliant young experim entalists, Delaroche and B erard, men whose names arc still cited in contem porary scientific litera ture. Laplace immediately made use of these m easurem ents in an indirect theoretical com putation of the speed of sound in air, and the discrepancy between theory and measurement dropped from 20 per cent to 2.5 per cent, a recognized trium ph in view of the state of m easurem ent.2 B ut today no one can explain how' this trium ph can have occurred. Laplaces interpretation of Delaroche and B crards figures made use of the caloric theory in a region where our own science is quite certain th a t th a t theory differs from directly relevant quantitative experiment by about 40 per cent. T here is, how ever, also a 12 per cent discrepancy between the m easurements of Delaroche and Berard and the results of equivalent experim ents today. We are no longer able to get their quantitative result. Yet, in Laplaces perfectly straightforw ard and essential com putation from the theory, these two discrepancies, experi mental and theoretical, cancelled to give close final agreem ent between the pre dicted and measured speed of sound. We may not, I feel sure, dismiss this as the result of mere sloppiness. Both the theoretician and the experim entalists involved were men of the very highest caliber. R ather we m ust here see evi dence of the way in which theory and experim ent m ay guide each other in the exploration of areas new to both. These examples may enforce the point draw n initially from the examples in the last section. Exploring the agreem ent between theory and experiment into new areas or to new limits of precision is a difficult, unrem itting, and, for many, exciting job. Though its object is neither discovery nor confirmation, its appeal is quite sufficient to consume almost the entire time and attention of those physical scientists who do quantitative work. I t dem ands the very best of their imagination, intuition, and vigilance. In addition when combined w-ith those of the last section these examples may show something more. T hey may, th at is, indicate why new laws of nature are so very seldom discovered sim ply by inspecting the results of measurem ents made w ithout advance knowledge of those laws. Because most scientific laws have so few quantitative points of contact with nature, because investigations of those contact points usually dem and such laborious instrum entation and approxim ation, and because nature itself needs to be forced to yield the appropriate results, the route from theory
20 T . S. K uhn. T h e C aloric T h eo ry of A d iab atic C om pression, Isis. 1958, 49: 132-140.



o r law to measurement can alm ost never be travelled backw ards. N um bers gathered w ithout some knowledge of the regularity to be expected almost never speak for themselves. Almost certainly they remain ju st numbers. T his does not mean th at no one has ever discovered a q uantitative regularity merely by measuring. Boyles Law relating gas pressure with gas volume, H ookes Law relating spring distortion with applied force, and Joules relation ship between heat generated, electrical resistance, and electric current were all the direct results of m easurem ent. T here are other examples besides. But, partly just because they are so exceptional and p artly because they never occur until the scientist m easuring knows everything but the particular form of the quantitative result he will obtain, these exceptions show ju st how improbable quantitative discovery by quantitative measurem ent is. T h e cases of Galileo and D alton men who intuited a quantitative result as the sim plest expression of a qualitative conclusion and then fought n atu re to confirm it are very much the more typical scientific events. In fact, even Boyle did not find his Law until both he and two of his readers had suggested th at precisely th at law (th e simplest quantitative form th at yielded the observed qualitative regu larity) ought to result if the numerical results were recorded.21 H ere, too, the quantitative implications of a qualitative theory led the way. One more example may make clear a t least some of the prerequisites for this exceptional sort of discovery. T he experim ental search for a law or laws describing the variation with distance of the forces between m agnetized and between electrically charged bodies began in the seventeenth century and was actively pursued through the eighteenth. Yet only in the decades immediately preceding Coulom bs classic investigations of 1785 did measurement yield even an approxim ately unequivocal answer to these questions. W hat made the dif ference between success and failure seems to have been the belated assim ila tion of a lesson learned from a p art of N ew tonian theory. Simple force laws, like the inverse square law for gravitational attractio n , can generally be ex pected only between m athem atical points or bodies th a t approxim ate to them. T he more complex laws of attraction between gross bodies can be derived from the simpler law governing the attraction of points by summing all the forces between all the pairs of points in the two bodies. B ut these laws will seldom take a simple m athem atical form unless the distance between th e two bodies is large compared with the dimensions of the attractin g bodies them selves. U nder these circumstances the bodies will behave as points, and ex perim ent may reveal the resulting simple regularity. Consider only the historically simpler case of electrical attractions and re pulsions.28 D uring the first half of the eighteenth century when electrical forces were explained as the results of effluvia emitted by the entire charged body alm ost every experim ental investigation of the force law involved plac ing a charged body a measured distance below one pan of a balance and then
27 M arie Boas, R o b ert D oyle and S e v e n teenth-C entury C hem istry (C am bridge, E n g land, 1958), p. 44. 28 M uch relevant m aterial w ill be found in D uane R oller an d D uane H . D . R oller, T h e D ev e h e m e n t o f the Concept o f E lectric C harge: E le ctricity jrotr. the G reeks to Coulom b, H a r v ard C ase H isto rie s in E x p erim en tal Science, C ase 8 (C am bridge, M ass., 1954), and in W o lf, E ig h teen th C entury, pp. 239-250, 268271.



measuring the weight th a t had to be placed in the other pan to ju st overcome the attraction. W ith this arrangem ent of apparatus, th e attraction varies in no simple way with distance. Furtherm ore, the complex way in which it does vary depends critically upon the size and m aterial of the attracted pan. M any of the men who tried this technique therefore concluded by throwing up their hands; others suggested a variety of laws including both the inverse square and the inverse first power; measurement had proved totally equivocal. Yet it did not have to be so. W hat was needed and what was gradually acquired from more qualitative investigations during the middle decades of the century was a more N ew tonian approach to th e analysis of electrical and magnetic phenom ena.2" As this evolved, experim entalists increasingly sought not the attraction between bodies but th at between point poles and point charges. In th a t form the experimental problem was rapidly and unequivocally resolved. T his illustration shows once again how large an am ount of theory is needed before the results of measurement can be expected to m ake sense. B ut, and this is perhaps the main point, when th a t much theory is available, the law is very likely to have been guessed w ithout measurement. Coulombs result, in particular, seems to have surprised few scientists. Though his m easure m ents were necessary to produce a firm consensus about electrical and mag netic attractions they had to be done; science cannot survive on guesses m any practitioners had already concluded that the law of attraction and re pulsion m ust be inverse square. Some had done so by simple analagy to N ew tons gravitational law; others by a more elaborate theoretical argum ent; still others from equivocal data. Coulombs Law was very much in the a ir before its discoverer turned to the problem. If it had not been, Coulomb might not have been able to m ake nature yield it. [R epeated discussions of this Section indicate two respects in which my text may be misleading. Some readers take my argum ent to mean th at the com m itted scientist can make nature yield any measurem ents th a t he pleases. A few of these readers, and some others as well, also think my paper asserts th at for the development of science, experiment is of decidedly secondary im por tance when compared with theory. U ndoubtedly the fault is mine, but I intend to be making neither of these points. If w hat I have said is right, nature undoubtedly responds to the theoretical predispositions with which she is approached by th e m easuring scientist. But th at is not to say either that nature will respond to any theory at all or that she will ever respond very much. Reexamine, for a historically typical exam ple, the relationship between the caloric and dynam ical theory of heat. In their abstract structures and in the conceptual entities they presuppose, these two theories are quite different and, in fact, incom patible. But, during the years
t0 A fuller account would have to describe both the ea rlier and the la ter approaches as N ew tonian. T h e conception that electric force results from eflluvia is p artly C artesian but in th e eighteenth century its locus-clossicus w as the ae th er th e o ry developed in N ew tons O ptieks. Coulom bs approach and th a t of several of his contem poraries depends far m ore directly on th e m athem atical theory in N ew to n 's Principia. F o r th e differences between these books, th eir influence in the eighteenth cen tu ry , an d th e ir impact on th e developm ent of electrical theory, see I. 13. Cohen, F ra n klin and N civ tor.: A n In q u iry into Specu la tive N r a t onion E x p e rim en ta l Scien ce and F ra n k I ins IVork in E le ctricity a s an E xa m p le T h creo f (P h ilad elp h ia, 1956).



when the two vied for the allegiance of the scientific community, the theoreti cal predictions th at could be derived from them were very nearly th e same (see the reference cited in note 26). If they had not been, the caloric theory would never have been a widely accepted tool of professional research nor would it have succeeded in disclosing th e very problem s th a t made transition to the dynamical theory possible. I t follows th a t any measurement which, like th at of Delaroche and B erard, fit one of these theories m ust have very nearly fit the other, and it is only within the experimental spread covered by the phrase very nearly th at nature proved able to respond to the theoretical predisposition of the m easurer. T h a t response could not have occurred with any theory a t all. T here are logically possible theories of, say, heat th a t no sane scientist could ever have m ade n ature fit, and there are problems, mostly philosophical, th a t m ake it worth inventing and examining theories of th a t sort. But those are not our problems, because those merely conceivable theories are not among the op tions open to the practicing scientist. His concern is with theories th at seem to fit w hat is known about nature, and all these theories, however different their structure, will necessarily seem to yield very similar predictive results. If they can be distinguished a t all by m easurem ents, those m easurem ents will usually strain the limits of existing experimental techniques. Furtherm ore, within the limits imposed by those techniques, the numerical differences at issue will very often prove to be quite small. Only under these conditions and within these limits can one expect nature to respond to preconception. On the other hand, these conditions and limits arc ju st the ones typical in the his torical situation. If this much about my approach is clear, th e second possible misunder standing can be dealt with more easily. By insisting th a t a quite highly de veloped body of theory is ordinarily prerequisite to fruitful m easurem ent in the physical sciences, I may seem to have implied th at in these sciences theory must always lead experiment and th a t the latter has a t best a decidedly sec ondary role. But th a t implication depends upon identifying experim ent with m easurem ent, an identification I have already explicitly disavowed. It is only because significant quantitative comparison of theories with nature comes a t such a late stage in the development of a science th a t theory has seemed to have so decisive a lead. If we had been discussing the qualitative experim entation th a t dom inates the earlier developmental stages of a physical science and th a t continues to play a role later on, the balance would be quite different. Perhaps, even then, we would not wish to say th a t experiment is prior to theory (though experience surely is ), but we would certainly find vastly more sym m etry and continuity in the ongoing dialogue between the two. Only some of my conclusions about the role of measurement in physical sci ence can be readily extrapolated to experim entation a t large.] IV. E X T R A O R D IN A R Y M E A S U R E M E N T To this point I have restricted attention to the role of m easurement in the norm al practice of natural science, the sort of practice in which all scientists are mostly, and most scientists are always, engaged. B ut natural science also



displays abnorm al situations times when research projects go consistently astray and when no usual techniques seem quite to restore them and it is through these rare situations th at m easurem ent shows its greatest strengths. In particular, it is through abnorm al states of scientific research th a t m easure m ent comes occasionally to play a m ajor role in discovery and in confirmation. L et me first try to clarify w hat I mean by an abnorm al situation or by w hat I am elsewhere calling a crisis state.30 I have already indicated th at it is a response by some p a rt of the scientific com m unity to its awareness of an anom aly in the ordinarily concordant relationship between theory and ex perim ent. B ut it is not, let us be clear, a response called forth by any and every anomaly. As the preceding pages have shown, current scientific practice alw ays embraces countless discrepancies between theory and experiment. D u r ing the course of his career, every natural scientist again and again notices and passes b y qualitative and quantitative anomalies th a t ju st conceivably might, if pursued, have resulted in fundam ental discovery. Isolated discrep ancies with this potential occur so regularly th a t no scientist could bring his research problem s to a conclusion if he paused for m any of them. In any case, experience has repeatedly shown th a t, in overwhelming proportion, these dis crepancies disappear upon closer scrutiny. T hey may prove to be instrum en tal effects, or they may result from previously unnoticed approxim ations in the theory, or they may, simply and mysteriously, cease to occur when the experim ent is repeated under slightly different conditions. M ore often than not the efficient procedure is therefore to decide th a t the problem has gone sour, th at it presents hidden complexities, and th a t it is time to p u t it aside in favor of another. F ortunately or not, th a t is good scientific procedure. B ut anomalies are not always dismissed, and of course they should not be. I f the effect is particularly large when compared with well-established meas ures of reasonable agreem ent applicable to sim ilar problems, or if it seems to resemble other difficulties encountered repeatedly before, or if, for personal reasons, it intrigues the experim enter, then a special research project is likely to be dedicated to it.*1 At th a t point th e discrepancy will probably vanish through an adjustm ent of theory or apparatus; as we have seen, few anom a lies resist persistent effort for long. But it may resist, and. if it docs, we may have the beginning of a crisis or abnorm al situation affecting those in whose usual area of research the continuing discrepancy lies. They, at least, having exhausted all the usual recourses of approxim ation and instrum enta tion, may be forced to recognize th a t som ething has gone wrong, and their behavior as scientists will change accordingly. At this point, to a vastly greater extent than a t any other, the scientist will s ta rt to search at random , trying anything at all which he thinks may conceivably illuminate the n atu re of his difficulty. If th a t difficulty endures long enough, he and his colleagues may even begin to wonder w hether their entire approach to the now problem atic range of natural phenomena is not somehow askew.
30 See note 3. 31 A recent exam ple of th e factor? d e te rm ining p u rsu it of a n anom aly has been investigated by B ernard B arber an d Renee C. F ox, " T h e C ase of th e F lo p p y -E ared R ab b its: A n Instance o f Serendipity G ained and Serendipity L ost, A m cr. So c. R e v ., 1958, 64: 128-136.



This is, of course, an immensely condensed and schematic description. U n fortunately, it will have to remain so, for the anatom y of the crisis state in natural science is beyond the scope of this paper. I shall rem ark only that these crises vary greatly in scope: they may emerge and be resolved within the work of an individual; more often they will involve most of those engaged in a particular scientific specialty; occasionally they will engross most of the members of an entire scientific profession. But, however widespread their im pact, there arc only a few ways in which they may be resolved. Sometimes, as has often happened in chem istry and astronom y, more refined experimental techniques or a finer scrutiny of the theoretical approxim ations will eliminate the discrepancy entirely. On other occasions, though I think not often, a dis crepancy th at has repeatedly defied analysis is simply left as a known anom aly, encysted w ithin the body of more successful applications of the theory. N ew tons theoretical value for the speed of sound and the observed preces sion of M ercurys perihelion provide obvious examples of effects which, though since explained, rem ained in the scientific literature as known anomalies for half a century or more. But there are still other modes of resolution, and it is they which give crises in science their fundam ental im portance. Often crises are resolved by the discover)' of a new natural phenomenon; occasionally their resolution dem ands a fundam ental revision of existing theory. Obviously crisis is not a prerequisite for discovery in the natural sciences. We have already noticed th at some discoveries, like that of Boyles Law and of Coulombs Law, emerge naturally as a quantitative specification of w hat is qualitatively already known. M any other discoveries, more often qualitative than quantitative, result from prelim inary exploration with a new instrum ent, e.g., the telescope, b attery , or cyclotron. In addition, there arc the famous accidental discoveries, Galvani and the twitching frogs legs, Roentgen and X -rays, Becquerel and the fogged photographic plates. T h e last two categories of discovery arc not, however, always independent of crises. I t is probably the ability to recognize a significant anom aly against th e background of cu r rent theory th a t most distinguishes the successful victim of an accident from those of his contem poraries who passed the same phenomenon by. (Is this not p art of the sense of P a steu rs famous phrase, In the fields of observation, chance favors only the prepared m ind ?)*2 In addition, the new instrum ental techniques that multiply discoveries are often themselves by-products of crises. V oltas invention of the b attery wras, for example, the outcome of a long a t tem pt to assim ilate Galvanis observations of frogs legs to existing electrical theory. And, over and above these somewhat questionable cases, there are a large number of discoveries th a t arc quite clearly the outcome of prior crises T he discovery of the planet N eptune was the product of an effort to account for known anomalies in the orbit of U ranus.35 T he nature of both chlorine and carbon monoxide was discovered through attem pts to reconcile Lavoisiers new chem istry with observation.54 T he so-called noble gases were the prod32 F ro m P a ste u rs inaugural address a t Lille in 1854 as quoted in R ene V allery-R adot, L a V ie de P asteur ( P a ris , 1903), p. 88. 83 A ngus A rm itag e, A C entury o f A stro n o m y (London, 1950), pp. 111-115. 51 F o r chlorine see E rn s t von M eyer. A H isto ry o f C h em istry fro m the E a rliest T u n es to the P resen t D a y, tran s. G. M G ow an (L o n don. 1891), pp. 224-227. F o r carbon m onoxide see J . R . P artin g to n , A S h o r t H isto r y o f



ucts of a long series of investigations initiated by a small b u t persistent anom aly in the measured density of nitrogen.54 T he electron was posited to explain some anomalous properties of electrical conduction through gases, and its spin was suggested to account for other sorts of anomalies observed in atom ic spec tra.'1 '1 T h e discovery of the neutrino presents still another example, and the list could be extended.37 I am not certain how large these discoveries-through-anom aly would rank in a statistical survey of discovery in the natural sciences.3* T hey are, how ever, certainly im portant, and they require disproportionate emphasis in this paper. To the extent that measurement and q uantitative technique play an especially significant role in scientific discovery, they do so precisely because, by displaying serious anomaly, they tell scientists when and where to look for a new qualitative phenomenon. T o the n atu re of th a t phenomenon, they usu ally provide no clues. W hen m easurem ent departs from theory, it is likely to yield mere num bers, and their very neutrality makes them particularly sterile as a source of remedial suggestions. But num bers register the departure from theory with an authority and finesse th a t no qualitative technique can dupli cate, and th a t departure is often enough to sta rt a search. N eptune might, like U ranus, have been discovered through an accidental observation; it had, in fact, been noticed by a few earlier observers who had taken it for a pre viously unobserved star. W hat was needed to draw attention to it and to make its discovery as nearly inevitable as historical events can be was its involve ment, as a source of trouble, in existing q uantitative observation and existing theory. I t is hard to see how either electron-spin or the neutrino could have been discovered in any other way. T he case both for crises and for measurement becomes vastly stronger as soon as we turn from the discovery of new natural phenom ena to the inven tion of fundam ental new theories. Though the sources of individual theoreti cal inspiration may be inscrutable (certainly they m ust remain so for this pa}>er), the conditions under which inspiration occurs is not. 1 know of no fundam ental theoretical innovation in natural science whose enunciation has not been preceded by clear recognition, often common to most of the profes sion, th at something was the m atter with the theory then in vogue. T h e state
C hem istry, 2nd cd. (L ondon, 1948), pp. 140141; and J. R. P artin g to n and D. M cK ic, "H isto rica l S tudies of the P h logiston T h e o ry : IV . L ast P hases of the T h eo ry , A n n a ls o f Science, 1939, 4: 365. 3S See note 7. w F o r useful surveys of th e experim ents w hich led to the discovery of th e electron sec T . W . Chalm ers, H isto ric R esearches: Chapters in the H isto r y o f P hysical and C hem ical D iscovery (L ondon, 1949), pp. 187-217, and J. J . T hom son, R ecollections and R eflection s (N e w Y ork, 1937), pp. 325-371. F o r electronspin sec F. K. R ichtm eycr, E. H . K cn n ard , aud T . L auritscn, Introduction to M odern P hysics, 5th ed. (N e w Y ork, 1955), p. 212. 87 R o g ers D . R usk, Introduction to A to m ic an d N u c le a r P h ysics (N e w Y ork, 1958), pp. 328-330. I kn o w of no o th e r elem entary account recent enough to include a description of the physical detection of th e neutrino. Because scientific atten tio n is often concen trated upon problem s th a t sccni to display anom aly, th e prevalence of d iscovery-throughanom aly m ay be one reason for the prevalence of sim ultaneous discovery in the sciences. F o r evidence th a t it is not the only one see T . S. K uhn, C onservation of E n erg y a s a n E xam pie of Sim ultaneous D iscovery, C ritical P roblem s in the H isto r y o f Science, ed. M arshall C lag ctt (M adison, 1959), pp. 321-356, but notice th a t m uch of w hat is th e re said about the em ergence of conversion processes' also dcscribes th e evolution of a crisis state.



of Ptolem aic astronom y was a scandal before Copernicus announcem ent.30 Both Galileos and N ew tons contributions to th e study of motion were ini tially focused upon difficulties discovered in ancient and medieval theory.40 N ew tons new theory of light and color originated in the discovery th a t exist ing theory would not account for the length of the spectrum , and the wave theory th a t replaced N ewtons was announced in the midst of growing concern about anomalies in the relation of diffraction and polarization to N ew tons theory.4 1 Lavoisiers new chem istry was born after the observation of anom a lous weight relations in combustion; thermodynam ics from th e collision of two existing nineteenth-century physical theories; quantum mechanics from a variety of difficulties surrounding black-body radiation, specific heat, and the photoelectric effect.42 Furtherm ore, though this is not the place to show it, each of these difficulties, except the optical one observed by N ewton, had been a source of concern before (b u t usually not long before) the theory th at re solved it was announced. I suggest, therefore, th at though a crisis or an abnorm al situation is only
39 K uhn, C opem ican R evolution, pp. 138140. 270-271; A. R. H all. T h e Scientific R e v o lution. 1S00-1S00 (L ondon, 1954), pp. 13-17. N o te p articu la rly th e role of ag itatio n for calendar reform in intensifying th e crisis. 40 K uhn, C opem ican R evo lu tio n , pp. 237260, and item s in bibliography on pp. 290-291. 41 F o r N ew ton see T . S. K uhn, "N ew to n s O ptical P apers," in Isaac N e iv ton's P apers & L e tte rs on N a tu ra l P hilosophy, cd. I. B. Cohen (C am bridge, M ass., 1958), pp. 27-45. F o r the w ave theory sec E . T . W h itta k er, H isto r y o f the T heories o f A e th e r and E lectricity, T he Classical T heories, 2nd cd. (I^ondon, 1951), pp. 94-109, and W hcw cll, Inductive Sciences, I I , 396-466. T h ese references clearly delineate th e crisis that characterized optics w hen F resncl independently began to develop the w ave theory a fte r 1812. But they say too little about eighteenth-century developm ents to indicate a crisis p rio r to Y oungs e a rlie r defense of the w ave theory in and a fte r 1801. In fact, it is not alto g eth er clear tin t th e re w as one, o r at least tliat th e re w as a new one. N ew tons c o r puscular theory of light had never been quite universally acccpted, and Y oung's early oppo sition to it w as based entirely upon anom alies tliat liad been g enerally recognized and often exploited before. W e m ay need to conclude th a t m ost of the eighteenth century w as c h a r acterized by a low -level crisis in optics, for the dom inant theory w as never im m une to fundam ental criticism and attack . T h a t w ould be sufficient to m ake the point that is of concern here, but I suspect a careful Study of the eighteenth-century optical lite ra tu re w ill perm it a still stro n g er conclusion. A cursory look a t th a t body of lite ra tu re sug gests th a t the anom alies of N ew tonian optics w ere fa r m ore ap p aren t and pressing in the tw o decades before Y oung's w ork than they had been before. D uring the 1780s the av a ila bility of ach ro m atic lenses and prism s led to num erous proposals fo r an astronom ical d e te r m ination of th e relativ e m otion of the sun and stars. (T h e references in W h itta k e r, op. cit., p. 109, lead d irectly to a fa r la rg e r lite ra tu re .) B ut these all depended upon lig h t's moving ntore quickly in glass than in a ir and thus gav e new relevance to an old controversy. L A b b i H a u v dem onstrated experim entally (M e m . d e IA cad. (1788), pp. 34-60) th a t H u y ghcns w ave-theoretical trea tm e n t of double refraction had yielded b etter resu lts than N ew tons co rp u scu lar treatm en t. T h e resulting problem leads to the prize offered by the F ren ch Academ y in 1808 and thus to M alus discovery of polarization by rcflcction in the sam e year. O r again, the P hilosophical T ra n s actions for 1796, 1797, and 1798 contain a series of tw o article s by B rougltam and a third by P rcv o st w hich show still o th er difficulties in N ew to n s theory. A ccording to P rcv o st, in p articu lar, the so rts of forces w hich m ust be ex erted on lig h t a t an interface in o rd er to explain reflection and refraction a rc not com patible w ith th e so rts of forces needed to e x plain inflection (P h il. T rans., 1798, 84: 325328. B io g rap h ers of Y oung m ight pay m ore atten tio n th an they hav e to the tw o B rougham p apers in the preceding volum es. T h ese d is play an intellectual com m itm ent th a t goes a long w ay to explain B rougham 's subsequent v itrio lic attac k upon Y oung in the pages of the E d inburgh R eview .) 4* R ichtm eyer el a!., M o d em P hysics, pp. 89-94, 124-132, an d 409-414. A m ore ele m en tary account of th e black-body problem and of th e photoelectric effect is included in G erald H olton, Introduction to Concepts and T h eo ries in P hysical Science (C am bridge, M ass., 1953), pp. 528-545.



one of the routes to discovery in the natural sciences, it is prerequisite to fundam ental inventions of theory. Furtherm ore, I suspect th a t in th e crea tion of the particularly deep crisis th a t usually precedes theoretical innova tion, m easurem ent makes one of its two most significant contributions to scientific advance. M ost of the anomalies isolated in the preceding paragraph were quantitative or had a significant q uantitative component, and, though the subject again carries us beyond the bounds of this essay, there is excellent reason why this should have been the case. Unlike discoveries of new natural phenom ena, innovations in scientific th e ory are not simply additions to the sum of what is already known. Almost always (alw ays, in the m ature sciences) the acceptance of a new theory de mands the rejection of an older one. In the realm of theory, innovation is thus necessarily destructive as well as constructive. B ut, as the preceding pages have repeatedly indicated, theories are, even more than laboratory in strum ents, the essential tools of the scientists trade. W ithout their constant assistance, even the observations and m easurem ents made by the scientist would scarcely be scientific. A threat to theory is therefore a threat to the scientific life, and, though the scientific enterprise progresses through such threats, the individual scientist ignores them while he can. P articularly, he ignores them if his own prior practice has already committed him to th e use of the threatened theory.44 I t follows that new theoretical suggestions, de structive of old practices, rarely if ever emerge in the absence of a crisis th at can no longer be suppressed. N o crisis is, however, so hard to suppress as one th at derives from a qu an ti tative anom aly th at has resisted all the usual efforts a t reconciliation. Once the relevant measurements have been stabilized and the theoretical approxi mations fully investigated, a quantitative discrepancy proves persistently ob trusive to a degree th at few qualitative anomalies can m atch. By their very nature, qualitative anomalies usually suggest ad hoc modifications of theory th a t will disguise them, and once these modifications have been suggested there is little way of telling w hether they are good enough. An established quantitative anomaly, in contrast, usually suggests nothing except trouble, but a t its best it provides a razor-sharp instrum ent for judging the adequacy of proposed solutions. K epler provides a brilliant case in point. A fter prolonged struggle to rid astronom y of pronounced quantitative anom alies in the motion of M ars, he invented a theory accurate to 8' of arc, a measure of agreement th a t would have astounded and delighted any astronom er who did not have access to the brilliant observations of Tycho Brahe. But from long experience K epler knew B rahe's observations to be accurate to 4' of arc. T o us, he said, Divine goodness has given a most diligent observer in Tycho Brahe, and it is therefore right th at we should with a grateful mind make use of this gift to find the true celestial motions. K epler next attem pted com putations with non
43 E vidence for this cffcct of p rio r experience w ith a theory is provided by th e w ellknow n, but inadequately investigated, youthfulness o f fam ous innovators a s well a s by the w ay in w hich younger men tend to clu ster to the new er theory. P lan ck s statem ent about the la tte r phenom enon needs n o citation. A n e a rlie r and p articu larly m oving version o f the sam e sentim ent is provided by D arw in in the last ch ap ter of T h e O ri(/in o f Species. (S e c th e 6th ed. [N ew Y ork, 1889J, I I , 295-296.)



circular figures. T h e outcome of those trials was his first two Laws of plane ta ry motion, the Laws th a t for the first tim e made the Copernican system work.4* Two brief examples should make clear the differential effectiveness of quali tative and quantitative anomalies. Newton was apparently led to his new the ory of light and color by observing the surprising elongation of the solar spectrum. Opponents of his new theory quickly pointed out that the exist ence of elongation had been known before and th a t it could be treated by existing theory. Q ualitatively they were quite right. B ut utilizing Snells quantitative law of refraction (a law th a t had been available to scientists for less than three decades), Newton was able to show th a t th e elongation pre dicted by existing theory was quantitatively far smaller than the one observed. On this quantitative discrepancy, all previous qualitative explanations of elon gation broke down. Given the quantitative law of refraction, N ew tons ulti mate, and in this case quite rapid, victory was assured.45 T he developm ent of chemistry provides a second striking illustration. I t was well known, long be fore Lavoisier, th a t some m etals gain weight when they arc calcined (i.e., roasted). Furtherm ore, by the middle of the eighteenth century this qualita tive observation was recognized to be incompatible with at least the sim plest versions of the phlogiston theory, a theory th a t said phlogiston escaped from the metal during calcination. B ut so long as the discrepancy remained quali tative, it could be disposed of in numerous w ays: perhaps phlogiston had negative weight, or perhaps fire particles lodged in the roasted metal. There were other suggestions besides, and together they served to reduce the urgency of the qualitative problem. T h e developm ent of pneum atic techniques, how ever, transform ed the qualitative anom aly into a q uantitative one. In the hands of Lavoisier, they showed how much weight was gained and where it came from. T hese were d ata with which the earlier qualitative theories could not deal. Though phlogistons adherents gave vehem ent and skillful battle, and though their qualitative argum ents were fairly persuasive, the q u an tita tive argum ents for Lavoisiers theory proved overwhelming.4 These examples were introduced to illustrate how difficult it is to explain away established quantitative anomalies, and to show how much more effec tive these are than qualitative anomalies in establishing uncvadablc scientific crises. B ut the examples also show something more. T hey indicate th at meas urem ent can be an immensely powerful weapon in the b attle between two
44 J . L. E . D rcycr, A H isto r y o f A stro n o m y fro m T hales to K epler, 2nd cd. (N e w Y ork, 1953), pp. 385-393. 46 K uhn, N ew tons O ptical P apers, pp. 31-36. 46 T h is is a slig h t oversim plification, since the Iwttle between I-avoisier's new chem istry an d its opponents really im plicated m ore tlan com bustion processes, and the full range of relevant evidence cannot be trea ted in term s of com bustion alone. U seful elem entary accounts of L avoisiers contributions can be found in : J . B. C onant, T h e O ve rth ro w o f the P h lo g iston T h eo ry , H a rv a rd C ase H isto rie s in E x peri m ental Scicncc, Case 2 (C am bridge, M ass., 1950), an d D . M cK ic, A n to in e iM Voisicr: S c ie n tist, E conom ist, S o cia l R e fo rm e r (N ew Y ork, 1952). M aurice D aum as, L a vo isier, T heoricien e t e x p trim e n tc u r ( P a ris , 1955) is the best recent scholarly review . J . H . W h ite, T h e PMogisto>v T h e o ry (L o n d o n , 1932) and especially J . R . P artin g to n and D. M cK ie, "H isto ric a l S tudies of the P hlogiston T h e o ry : IV . L ast P h ases of the T h eo ry , A n n a ls of Scien ce, 1939, 4: 113-149, give m ost detail ab o u t th e conflict betw een th e new theory and th e old.



theories, and that, I think, is its second particularly significant function. F u r therm ore, it is for this functionaid in the choice between theories and for it alone, th at we must reserve the word confirm ation. W e must, th a t is, if confirm ation is intended to denote a procedure anything like w hat scientists ever do. T h e m easurem ents th at display an anom aly and thus create crisis may tem pt the scientist to leave science or to tran sfer his attention to some other p a rt of the field. B ut, if he stays where he is, anomalous observations, quantitative or qualitative, cannot tem pt him to abandon his theory until an other one is suggested to replace it. Ju st as a carpenter, while he retains his craft, cannot discard his toolbox because it contains no ham m er fit to drive a particular nail, so the practitioner of science cannot discard established th e ory because of a felt inadequacy. A t least he cannot do so until shown some other way to do his job. In scientific practice the real confirmation questions always involve the comparison of two theories with each other and with the world, not the comparison of a single theory w ith the world. In these threeway comparisons, m easurem ent has a particular advantage. T o sec where m easurem ents advantage resides, I m ust once more step briefly, and hence dogmatically, beyond the bounds of this essay. In the tra n sition from an earlier to a later theory, there is very often a loss as well as a gain of explanatory power.47 N ew tons theory of planetary and projectile mo tion was fought vehemently for more than a generation because, unlike its main com petitors, it demanded the introduction of an inexplicable force th at acted directly upon bodies a t a distance. C artesian theory, for example, had attem pted to explain gravity in terms of the direct collisions between elemen tary particles. T o accept Newton m eant to abandon the possibility of any such explanation, or so it seemed to most of N ew tons immediate successors.48 Similarly, though the historical detail is more equivocal, Lavoisiers new chemi cal theory was opposed by a num ber of men who felt th a t it deprived chem istry of one principal traditional function the explanation of the qualitative properties of bodies in term s of the particular combination of chemical p rin ciples th at composed them .40 In each case the new theory was victorious, but the price of victory was the abandonm ent of an old and partly achieved goal. F or eighteentli-century N ewtonians it gradually became unscientific to ask for the cause of gravity; nineteenth-century chemists increasingly ceased to ask for the causes of particular qualities. Yet subsequent experience has shown th a t there was nothing intrinsically unscientific about these questions. Gen
*1 T h is point is ccn tral to the reference citcd in note 3. In fact, it is largely th e ncccssity of balancing gains and losses and the controvcrsies th a t so often resu lt from disagreem ents about an appro p riate balancc th a t m ake it ap p ro p riate to describe changes of theory a s revolutions.0 48 Cohen, F ra n klin and N e w to n , C h ap ter 4 ; P ierre B runet, [ .introduction d es theories de N e w to n en F rance au . m ? siecle ( P a ris , 1931). 49 O n this trad itio n al task of chem istry sec E. Meyer.son. Id en tity and R ea lity, tran s. K. L ow cnberg (L ondon, 1930), C hapter X , p ar-

tic u la rly pp. 331-336. M uch essential m aterial is also scattered th ro u g h H fenc M etzger, L es .rw'i < i la fin d tt xviii* sicete, vol. I (P a ris , 1923). an d NeitiOH, S ta h l. Poerltaave, et la doctrine chim iquc ( P a ris . 1930). N otice particu larly th a t the phlogistonists. w ho looked upon o res a s elem en tary bodies from which the m etals w ere com pounded by addition of phlogiston, could explain w hy the m etals w ere so m uch m ore like each o th e r th an w ere the ores fro m w hich they w ere com pounded. A ll m etals had a principle, phlogiston, in common. N o such explanation w as possible on Lavoisiers theory.



eral relativity does explain gravitational attraction, and quantum mechanics does explain many of the qualitative characteristics of bodies. W e now know w hat makes some bodies yellow and others transparent, etc. B ut in gaining this immensely im portant understanding, we have had to regress, in certain respects, to an older set of notions about the bounds of scientific inquiry. Problem s and solutions th at had to be abandoned in em bracing classic theo ries of modern science are again very much w ith us. T he study of the confirmation procedures as they are practiced in the sci ences is therefore often the study of w hat scientists will and will not give up in order to gain other particular advantages. T h a t problem has scarcely even been stated before, and I can therefore scarcely guess w hat its fuller investi gation would reveal. But impressionistic study strongly suggests one signifi cant conclusion. I know of no case in the development of science which exhibits a loss of quantitative accuracy as a consequence of the transition from an earlier to a later theory. N or can I imagine a debate between scien tists in which, however hot the emotions, th e search for greater numerical accuracy in a previously quantified field would be called unscientific. Prob ably for the same reasons th a t make them particularly effective in creating scientific crises, the comparison of numerical predictions, where th ey have been available, has proved particularly successful in bringing scientific con troversies to a close. W hatever the price in redefinitions of science, its m eth ods, and its goals, scientists have shown themselves consistently unwilling to compromise the numerical success of their theories. Presum ably there are other such desiderata as well, but one suspects that, in case of conflict, meas urem ent would be the consistent victor. V. M E A S U R E M E N T IN T H E D E V E L O P M E N T O F PH Y SIC A L S C IE N C E T o this point we have taken for granted th a t m easurem ent did play a cen tral role in physical science and have asked about the n atu re of th at role and the reasons for its peculiar efficacy. Now we must ask, though too late to anticipate a com parably full response, about the way in which physical sci ence came to make use of quantitative techniques at all. T o m ake th a t large and factual question manageable, I select for discussion only those p a rts of an answer which relate particularly closely to w hat has already been said. One recurrent implication of the preceding discussion is that much qualita tive research, both empirical and thcorctical, is normally prerequisite to fruit ful quantification of a given research field. In th e absence of such prior work, the methodological directive, Go ye forth and measure, may well prove only an invitation to w aste time. I f doubts about this point remain, they should be quickly resolved by a brief review of the role played by q u antitative tech niques in the emergence of the various physical sciences. L et me begin by asking w hat role such techniques had in the scientific revolution th at centered in the seventeenth century. Since any answer m ust now be schematic, I begin by dividing the fields of physical science studied during the seventeenth century into two groups. T he



first, to be labeled the traditional sciences, consists of astronom y, optics, and mechanics, all of them fields th a t had received considerable qualitative and quantitative development in antiquity and during the M iddle Ages. These fields are to be contrasted with w hat I shall call the Baconian sciences, a new cluster of research areas th a t owed their status as scicticcs to the seventeenth centurys characteristic insistence upon experim entation and upon the com pilation of natural histories, including histories of the crafts. T o this second group belong particularly the study of heat, of electricity, of magnetism, and of chemistry. Only chem istry had been much explored before the Scientific Revolution, and the men who explored it had almost all been either craftsm en or alchemists. If we except a few of the a r ts Islam ic practitioners, the em er gence of a rational and system atic chemical tradition cannot be dated earlier than the late sixteenth century.30 M agnetism , heat, and electricity emerged still more slowly as independent subjects for learned study. Even more clearly than chem istry, they are novel by-products of the Baconian elem ents in the new philosophy.M T he separation of traditional from Baconian sciences provides an im portant analytic tool, because the man who looks to the Scientific Revolution for ex amples of productive m easurem ent in physical science will find them only in the sciences of the first group. F urther, and perhaps more revealing, even in these traditional sciences m easurem ent was most often effective ju s t when it could be perform ed with well-known instrum ents and applied to very nearly traditional concepts. In astronom y, for example, it was Tycho B rahes en larged and better-calibrated version of medieval instrum ents th a t made the decisive quantitative contribution. T h e telescope, a characteristic novelty of the seventeenth century, was scarcely used quantitatively until the last third of the century, and that quantitative use had no effect on astronom ical theory until B radleys discovery of aberration in 1729. Even th at discovery was iso lated. Only during the second half of the eighteenth century did astronom y begin to experience the full effects of the immense improvements in q u an ti tative observation that the telescope perm itted. O r again, as previously in dicated, the novel inclined plane experim ents of the seventeenth century were not nearly accurate enough to have alone been the source of the law of uni form acceleration. W hat is im portant about them and they are critically im portant is the conception th a t such m easurem ents could have relevance to the problems of free fall and of projectile motion. T h a t conception implies a fundamental shift in both the idea of motion and the techniques relevant to its analysis. But clearly no such conception could have evolved as it did if
60 Boas, R o b rrt Hoyle, pp. 48-66. 51 F o r clcctricity see, R o ller and R oller, Concept o f E lectric C harge. H a rv a rd C ase H isto rie s in E xperim ental Science, C ase 8 (C am bridpc, M ass.. 1954), and. E d g a r Zilscl, " T h e O rig in s of W illiam G ilbert's Scicntific M ethod, J. H ist. Ideas. 1941, 2: 1-32. I a g re e w ith those w ho feel Zilscl ex ag g erates the im portance of a single factor in the genesis of clcctrical science and. by im plication, o f Baconianism , but the cra ft influences he describes cannot conceivably be dism issed. T h e re is no equally satisfactory discussion of th e dcvclopm cn t of th erm al science before th e eighteenth ccntury. but W olf, 16th a n d 17th C enturies. pp. 82-92 and 275-281 w ill illu strate the tran sform ation produced by Baconianism . 52 W olf, E ig h teen th C entury, pp. 102-145, an d W hcw ell, In d u ctive S cien ces. pp. 213-371. P a rtic u la rly in th e latter, notice the difficulty in sep aratin g advances due to im proved instru m cn tatio n from those due to im proved theory. T h is difficulty is not due p rim arily to W hcw ell's m ode of presentation.



m any of the subsidiary concepts needed for its exploitation had not existed, a t least as developed embryos, in the works of Archimedes and of the scho lastic analysts of motion.53 Here again the effectiveness of q uantitative work depended upon a long-standing prior tradition. Perhaps the best test case is provided by optics, the third of my traditional sciences. In this field during the seventeenth century, real q uantitative work was done with both new and old instrum ents, and the work done with old instrum ents on well-known phenomena proved the more im portant. T he Sci entific Revolutions reform ulation of optical theory turned upon N ew tons prism experiments, and for these there was much qualitative precedent. N ew tons innovation was the quantitative analysis of a well-known qualitative ef fect, and th at analysis was possibly only because of the discovery, a few decades before N ew tons work, of Snells law of refraction. T h a t law is the vital quantitative novelty in the optics of the seventeenth century. I t was, however, a law th a t had been sought by a scries of brilliant investigators since the time of Ptolemy, and all had used ap p aratu s quite similar to that which Snell employed. In short, the research which led to N ew tons new theory of light and color was of an essentially traditional nature.44 M uch in seventeenth-century optics was, however, by no means traditional. Interference, diffraction, and double refraction were all first discovered in the half-century before N ew tons O pticks appeared; all were totally unexpected phenom ena; and all were known to N ew ton. On two of them Newton con ducted careful quantitative investigations. Yet the real impact of these novel phenom ena upon optical theory was scarcely felt until the work of Young and Fresnel a century later. Though Newton was able to develop a brilliant pre liminary theory for interference effects, neither he nor his immediate succes sors even noted th at that theory agreed with quantitative experim ent only for the limited case of perpendicular incidence. N ew tons measurements of dif fraction produced only the most qualitative theory, and on double refraction he seems not even to have attem pted q uantitative work of his own. Both Newton and H uyghen announced m athem atical laws governing the refraction of the extraordinary ray, and the latter showed how to account for this be havior by considering the expansion of a spheroidal wave front. But both m athem atical discussions involved large extrapolations from scattered q u an ti tative data of doubtful accuracy. And almost a hundred years elapsed before quantitative experiments proved able to distinguish between these two quite different m athem atical formulations.5* As with the other optical phenomena
a* F o r pre-G alilean w o rk sec, M arshall C lagett, T h e Science o f M echanics in the M iddle A g e s (M adison, W is., 1959), p articu larly P a rts II & I I I . F o r G alileos use of this w ork see, A lexandre K oyrc. t.tu d es Calileennes, 3 vols. ( P a ris , 1939), p articu larly I & II. f * A . C. Crom bie, A u g u stin e to Galileo (L ondon, 1952), pp. 70-82, and W olf, 16th & 17th C enturies, pp. 244-254. 56 Ibid., pp. 254-264. M F o r the seventeenth-century w ork (in eluding H uyghens geom etric construction) see the rcfercncc in (he preceding: note. T he cightccnth-ccntury investigations of these phenom ena have scarcely been studied, but for w hat is know n see, Joseph P riestley , H isto ry . . . o f D iscoveries relating to V isio n . L ig h t, and C olours (L ondon. 1772), pp. 279-316, 498520, 548-562. T h e ea rliest exam ples I know of m ore precise w ork on double refraction are, R . J . H atiy , " S u r la double refractio n du S p ath d Islandc, M em . d I'A cad. (1788), pp. 34-61, an d , W . H . W o llasto n , "O n the oblique R efractio n of Iceland C ry stal, P h il. Trans., 1802, 92: 381-386.



discovered during the Scientific Revolution, m ost of th e eighteenth century was needed for the additional exploration and instrum entation prerequisite to quantitative exploitation. T urning now to the Baconian sciences, which throughout the Scientific Revolution possessed few old instrum ents and even fewer well-wrought con cepts, we find quantification proceeding even more slowly. Though the seven teenth century saw m any new instrum ents, of which a num ber were q u an ti tative and others potentially so, only the new barom eter disclosed significant quantitative regularities when applied to new fields of study. And even the barom eter is only an ap p aren t exception, for pneum atics, the field of its ap plication, was able to borrow en bloc the concepts of a far older field, hydro statics. As Toricelli put it, the barom eter m easured pressure at the bottom of an ocean of the clem ent air. 57 In the field of magnetism the only signifi cant seventeenth-century m easurem ents, those of declination and dip, were made with one or another modified version of the traditional compass, and these measurem ents did little to improve the understanding of magnetic phe nomena. For a more fundam ental quantification, magnetism, like electricity, awaited the work of Coulomb, Gauss, Poisson, and others in the late eight eenth and early nineteenth centuries. Before th at work could be done, a better qualitative understanding of attraction, repulsion, conduction, and other such phenom ena was needed. T h e instrum ents which produced a lasting quantifi cation had then to be designed with these initially qualitative conceptions in mind.5* Furtherm ore, the decades in which success was a t last achieved are alm ost the same ones th at produced the first effective contacts between meas urem ent and theory in the study of chem istry and of heat.59 Successful q u an tification of the Baconian sciences had scarcely begun before the last third of the eighteenth century and only realized its full potential in the nineteenth. T h a t realization exemplified in the work of Fourier, Clausius, Kelvin, and Maxwell is one facet of a second scientific revolution no less consequential than the seventeenth-century revolution. Only in the nineteenth century did the Baconian physical sciences undergo the transform ation which the group of traditional sciences had experienced two or more centuries before. Since Professor G uerlacs paper is devoted to chem istry and sincc I have al ready sketched some of the bars to quantification of electrical and magnetic phenom ena, I tak e my single more extended illustration from the study of heat. U nfortunately, much of the research upon which such a sketch should be based rem ains to be done. W h at follows is necessarily more tentative than w hat has gone before. M any of the early experim ents involving therm om eters read like investi gations o f th at new instrum ent rather than like investigations w ith it. How
67 See I.H .B . and A .G .H . S piers, T h e P h y s ical T reatises o f Pascal (N e w Y ork, 1937), p. 164. T h is w hole volum e displays the w ay in w hich seventeenth-century pneum atics took over concepts from hydrostatics. 6S F o r the quantification and early m athem atization of electrical science, see: R o ller and R oller, Concept o f E lectric C harge, pp. 66-80; W h itta k e r, A e th e r and E lectricity, pp. 53-66; and W . C. W a lk e r, T h e D etection and E stim atio n of E le ctric C h a rg e in the E ig h teenth C entury, A n n a ls o f Science, 1936, 1: 66-100. 69 F o r heat sec, D ouglas M cK ie an d N . II. de V . H cathcotc, T h e D iscovery o f Specific and L a ten t H e a ts (L ondon, 1935). In chem istry it m ay w ell be im possible to fix any d ate for th e first effective contacts betw een mcas-



could anything else have been the case during a period when it was totally unclear w hat the therm om eter measured? Its readings obviously depended upon the degree of heat but apparently in immensely complex ways. D e gree of heat had for a long time been defined by the senses, and the senses responded quite differently to bodies which produced th e same therm om etric readings. Before the thermometer could become unequivocally a laboratory instrum ent rath er than an experim ental subject, therm om etric reading had to be seen as the direct measure of degree of heat, and sensation had simul taneously to be viewed as a complex and equivocal phenomenon dependent upon a number of different parameters.** T h a t conceptual reorientation seems to have been completed in a t least a few scientific circles before the end of th e seventeenth century, b u t no rapid discovery of quantitative regularities followed. F irst scientists had to be forced to a bifurcation of degree of h eat into quantity of heat, on the one hand, and tem perature, on the other. In addition they had to select for close scrutiny, from the immense m ultitude of available thermal phenomena, the ones th a t could most readily be made to reveal q uantitative law. These proved to be: mixing two components of a single fluid initially a t different tem peratures, and radiant heating of two different fluids in identical vessels. Even when attention was focused upon these phenomena, however, scientists still did not get unequivocal or uniform results. As H eathcote and M cKie have brilliantly shown, the last stages in the developm ent of the concepts of specific and latent heat display intuited hypotheses constantly interacting with stubborn measurement, each forcing the other into line.*1 Still other sorts of work were required before the contributions of Laplace, Poisson, and Fourier could transform the study of therm al phenom ena into a branch of m athe matical physics.82 T his sort of pattern, reiterated both in the other Baconian sciences and in the extension of traditional sciences to new instrum ents and new phenomena, thus provides one additional illustration of this p ap ers most persistent thesis. T he road from scientific law to scientific m easurem ent can rarely be traveled
urcm ent and theory. V olum etric o r g ra v im ctric m easures w ere alw ays a n ingredient of chemical recipes and assays. By th e seventccnth century, fo r exam ple in the w ork of Boyle, w eight-gain o r loss w as often a clue to the thcorctical analysis of p articu la r reactions. B ut until the middle of the eighteenth century, the significance of chem ical m easurem ent seems alw ays to have been eith er dcscriptive (a s in recipes) or qualitative ( a s in dem onstrating a w ciglu-gain w ithout significant reference to its m ag n itu d e). O nly in the w o rk of B lack, L avoisier, and R ich ter docs m easurcm ent begin to play a fully quantitative role in the developm ent of chem ical law s and theories. F o r an introduction to these men and th e ir w ork see, J . B. P a rtin g to n , A S h o r t H isto ry o f C hem istry, 2nd cd. (L ondon, 1951), pp. 93-97, 122-128, and 161-163. 00 M aurice D aum as, L e s instru m en ts scientifiques a u x xvii* e t xviii* sitcle s ( P a ris , 1953), pp. 78-80, provides a n excellent b rief account of the slow stag es in the deploym ent of the therm om eter a s a scicntific instrum ent, R obert B oyle's N e w E xp e rim en ts and O bservations T o u ching C old illu strates th e sevcntccn th cen tu ry s need to d em onstrate th a t propcrly constructed therm om eters m u st rcplacc the senses in th erm al m easurem ents even though th e tw o give divergent results. Sec W o rks o f the H onourable R o b ert B o yle, cd. T . B irch, 5 vols. (L ondon, 1744), II, 240-243. 61 F o r th e elaboration of calo rim ctric conccpts sec, E . M ach, D ie P rincipien der W iirm elehre (L eipzig, 1919), pp. 153-181, and McK ie and H eathcote, Specific and L a ten t H eats. T h e discussion of K ra ffts w ork in the la tte r (pp. 59-63) provides a p articu larly strik in g exam ple of th e problem s in m aking m easurcm ent w ork. ** G aston B ach elard , E tu d e su r Invo lu tio n d u n problem e d e physique ( P a ris , 1928), and K uhn, "C alo ric T h eo ry of A diabatic C om prcssion.



in the reverse direction. T o discover quantitative regularity one must nor mally know w hat regularity one is seeking and ones instrum ents m ust be designed accordingly; even then nature may not yield consistent or generalizable results w ithout a struggle. So much for my m ajor thesis. T he preceding rem arks about the w ay in which quantification entered the modern physical sciences should, however, also recall this p ap ers minor thesis, for they re direct attention to the immense efficacy of quantitative experim entation u n dertaken within the context of a fully m athem atized theory. Sometime be tween 1800 and 1850 there was an im portant change in the character of research in many of the physical sciences, particularly in the cluster of re search fields known as physics. T h a t changc is w hat m akes me call the mathem atization of Baconian physical science one facet of a second scientific revolution. It would be absurd to pretend th a t m athem atization was more than a facet. The first half of the nineteenth century also witnessed a vast increase in the scale of the scientific enterprise, m ajor changes in p attern s of scientific o r ganization, and a total reconstruction of scientific education.63 But these changes affected all the sciences in much the same way. T hey ought not to explain the characteristics th a t differentiate the newly m athem atized sciences of the nineteenth century from other sciences of the same period. Though my sources are now impressionistic, I feel quite sure th at there are such charasteristics. Let me hazard the following prediction. Analytic, and in p art statistical, research would show th a t physicists, as a group, have displayed since about 1840 a greater ability to concentrate their attention on a few key areas of research than have their colleagues in less completely quantified fields. In the same period, if I am right, physicists would prove to have been more successful than most other scientists in decreasing the length of contro versies about scientific theories and in increasing the strength of the consensus th at emerged from such controversies. In short, I believe th at the nineteenthcentury m athem atization of physical science produced vastly refined profes sional criteria for problem selection and th at it simultaneously very much increased the effectiveness of professional verification p ro ced u res/4 These are, of course, just the changes that the discussion in Section IV would lead us to expect. A critical and com parative analysis of the development of physics during the p a st century-and-a-quarter should provide an acid test of those conclusions. Pending th a t test, can we conclude anything at all? I venture the following paradox: T he full and intim ate quantification of any science is a consumma tion devoutly to be wished. N evertheless, it is not a consummation th a t can effectively be sought by measuring. As in individual developm ent, so in the scientific group, m aturity comes most surely to those who know how to wait.
M S. F . M ason, M ain C urrents o f Scien tific T h o u g h t (N e w Y ork, 1956), pp. 352-363, p ro vides a n excellent brief sketch of these institutional changes. M uch additional m aterial is scattered through, J . T . M crz, H isto r y o f E uropean T hought in the N in eteen th C entury, vol. I (L ondon, 1923). c* F o r an exam ple of effective problem seIcction, note the esoteric q u an titativ e discrcpancics w hich isolated the th ree problem s photoelectric effect, black body radiation, and specific heats tliat gav e rise to quantum mechanics. F o r the new effectiveness of verification procedures, note th e speed w ith w hich this radical new th eo ry w as adopted by th e profession.



A P P E N D IX Reflecting on the other papers and on the discussion th a t continued through out the conference, two additional points th a t had reference to my own paper seem worth recording. U ndoubtedly there were others as well, but my mem ory has proved more than usually unreliable. Professor Price raised the first point, which gave rise to considerable discussion. The second followed from an aside by Professor Spengler, and I shall consider iLs consequences first. Professor Spengler expressed great interest in my concept of crises in the development of a science or of a scientific specialty, b u t added th a t he had had difficulty discovering more than one such episode in the developm ent of economics. T his raised for me the perennial, b u t perhaps not very im portant question about w hether or not the social sciences are really sciences a t all. Though I shall not even attem pt to answer it in th at form, a few further re m arks about the possible absence of crises in the developm ent of a social science may illuminate some p a rt of what is a t issue. As developed in Section IV, above, the concept of a crisis implies a prior unanim ity of the group th a t experiences one. Anomalies, by definition, exist only with respect to firmly established expectations. Experim ents can create a crisis by consistently going wrong only for a group that has previously ex perienced everythings seeming to go right. Now, as my Sections II and III should indicate quite fully, in the m ature physical sciences most things gen erally do go right. T he entire professional community can therefore ordinarily agree about the fundam ental concepts, tools, and problem s of its science. W ithout th at professional consensus, there would be no basis for the sort of puzzle-solving activity in which, as 1 have already urged, most physical scien tists are normally engaged. In the physical sciences disagreement about fun dam entals is, like the search for basic innovations, reserved lor periods of crisis.06 I t is, however, by no means equally clear th at a consensus of anything like similar strength and scope ordinarily characterizes the social scicnces. Experience with my university colleagues and a fortunate year spent at the C enter for Advanced Study in the Behavioral Sciences suggest th at the funda mental agreement which physicists, say, can norm ally take for granted has only recently begun to emerge in a few areas of social-science research. M ost other areas are still characterized by fundam ental disagreem ents about the definition of the field, its paradigm achievements, and its problems. While th a t situation obtains (as it did also in earlier periods of the developm ent of the various physical sciences), either there can be no crises or there can never be anything else. Professor P rices point was very different and far more historical. H e sug gested, and I think quite rightly, th at my historical epilogue failed to call attention to a very im portant change in the attitu d e of physical scientists to w ards measurement th a t occurred during the Scientific Revolution. In com4 5 1 have developed som e o th e r significant concom itants of th is professional consensus in m y paper, "T h e E ssential T en sio n : T ra d itio n an d Innovation in Scientific R esearch." T h a t paper ap p ears in, C alvin W . T a y lo r (c d .), T h e T h ird (1959) U n iversity o f U tah R esearch C onference on the Identification o f C reative S cien tific T a len t (U n iv ersity of U tah P re ss, 1959), pp. 162-177.


T H O M A S S. K U H N

meriting on D r. Crombies paper, Price had pointed out th a t not until the late sixteenth century did astronom ers begin to record continuous series of obser vations of planetary position. (Previously they had restricted themselves to occasional quantitative observations of special phenom ena.) Only in th at same late period, he continued, did astronom ers begin to be critical of their q u an titative data, recognizing, for example, th at a recorded celestial position is a clue to an astronomical fact rather than the fact itself. W hen discussing my paper, Professor Price pointed to still other signs of a change in the attitu d e towards measurement during the Scientific Revolution. For one thing, he emphasized, many more num bers were recorded. M ore im portant, perhaps, people like Boyle, when announcing laws derived from measurement, began for the first time to record their quantitative d ata, w hether or not th ey per fe ctly fit the law, rather than simply stating the law itself. I am somewhat doubtful th at this transition in attitu d e towards num bers proceeded quite so far in the seventeenth century as Professor Price seemed occasionally to imply. Hooke, for one example, did not re}>ort the num bers from which he derived his law of elasticity; no concept of significant figures seems to have emerged in the experim ental physical sciences before the nine teenth century. B ut I cannot doubt th at the change was in process and that it is very im portant. A t least in another sort of paper, it deserves detailed examination which I very much hope it will get. Pending th at exam ination, however, let me simply point out how very closely the development of the phenomena emphasized by Professor Price fits th e pattern I have already sketched in describing the effects of seventeenth-century Baconianism. In the first place, except perhaps in astronom y, the seventeenth-century change in attitu d e tow ards measurem ent looks very much like a response to the novelties of the methodological program of the new philosophy. Those novelties were not, as has so often been supposed, consequences of the belief th at observation and experiment were basic to science. As Crombie has bril liantly shown, th at belief and an accompanying methodological philosophy were highly developed during the M iddle Ages.* Instead, the novelties of method in the new philosophy included a belief that lots and lots of experi ments would be necessary (th e plea for natural histories) and an insistence th at all experim ents and observations be reported in full and naturalistic de tail, preferably accompanied by the names and credentials of witnesses. Both the increased frequency with which num bers were recorded and the decreased tendency to round them off are precisely congruent with those more general Baconian changes in the attitu d e towards experim entation at large. Furtherm ore, whether or not its source lies in Baconianism, the effective ness of the seventeenth-centurys new attitu d e towards num bers developed in very much the same way as the effectiveness of the other Baconian novelties discussed in my concluding section. In dynamics, as Professor K oyr6 has re peatedly shown, the new attitu d e had almost no effect before the later eight eenth century. T he other two traditional sciences, astronom y and optics, were affected sooner by the change, but only in their most nearly traditional parts.
See p articu larly h is R o b ert G rosseteste and the O rigins o f E xp e rim en ta l Science, 1100 1700 (O x fo rd , 1953).



And the Baconian sciences, heat, electricity, chem istry, etc., scarcely begin to profit from the new attitude until after 1750. Again it is in the work of Black, Lavoisier, Coulomb, and their contem poraries th a t the first truly significant effects of the change arc seen. And the full transform ation of physical scicncc due to th at change is scarcely visible before the work of Ampere, Fourier, Ohm, and Kelvin. Professor Price has, I think, isolated another very signifi cant seventeenth-century novelty. B ut like so m any of the other novel a tti tudes displayed by the new philosophy, the significant effects of this new attitude tow ards m easurem ent were scarcely m anifested in th e seventeenth century at all.