`
`3M COMPANY 2015
`Mylan Pharmaceuticals Inc. v. 3M Company
`IPR2015-02002
`
`
`
`THE NATURE AND HISTORY OF EXPERIMENTAL CONTROL
`
`By EDWIN G. BORING, Harvard University
`
`Solomon has discussed the history of the concept of control in experi-
`mental work with especial regard to the use of control groups in the design
`of experiments} finding no instance of the employment of a control group
`before the study of transfer by Thorndike and Woodworth in 19012 and
`no extensive use of control groups in experiments on transfer until Winch’s
`study in 1908.3 It is easily shown that the concept of control is basic to all
`experimental design and is,
`indeed,
`inherent in the essential relational
`nature of a fact. The purpose of this note is to analyze the concept more
`fully and to say something more about the history of both the concept and
`the word.
`
`The word control has three meanings: (1) a c/neck, in the sense of a veri-
`fication but thus also in the sense of a restraint, since verification restrains;
`
`in the sense of a checking and thus also in the sense of
`(2) a restraint,
`maintaining constancy; and (3) a guide or directing, in the sense of pro-
`ducing a precisely predetermined change, a constant and thus a restrained
`change. The word c/aecé itself has the first two meanings, though not the
`last, and the original meaning of control was c/oecté, for the word was
`counter-roll (contre-rolle), a duplicate register or account made to verify
`an official or first-made account and thus a check by a later roll upon the
`earlier. (Hence controller, which is misspelled comptroller because it has
`been thought of as meaning an accountant instead of a checker.) So the
`thought of correctness or conformity achieved by restraint runs all the way
`through the history of the word, even though what the psychologist hears
`nowadays about ‘controlling behavior’ suggests the promotion of action
`more than its restriction.4
`
`* Accepted for publication September 10, 1953.
`‘R. L. Solomon, An extension of control group design, Prycbol. Bull., 46, 1949,
`137-150.
`‘E. L. Thorndike and R. S. Woodworth, The influence of improvement in one
`mental function upon the efliciency of other functions, Ptyclaol. Rer/., 8, 1901, 247-
`261, 384-395, 553-564, esp. the eight-line paragraph on p. 558 where the use of a
`control group is described.
`’ W. H. Winch, The transfer of improvement of memory in school-children, Brit.
`I. Pry:/Jol., 2, 1908, 284-293. The article antedates the development of the statistical
`method for dealing with group differences.
`‘ Cf. B. F. Skinner, Science and Human Behavior, 1953, where the conception of
`control seems to me to be more positive than negative in the chapters entitled: The
`573
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`2
`
`
`
`574
`
`BORING
`
`The term control in the sense of a check or test observation or experi-
`ment came into scientific parlance in the latter half of the nineteenth
`century, as we shall see presently. By 1893 we find the New English
`Dictionary defining control as “a standard of comparison used to check
`the inferences deduced from an experiment by application of the Method
`of Difference,” which is the name of John Stuart Mill’s second method of
`experimental inquiry. The New English goes on to define control-experzl
`ment as "a test experiment with this end in view.” So control-experiment
`was definitely in the language before control group: had been thought of,
`and this concept takes us back to Mill’s four methods of experimental
`inquiry in his Logic of 1843.5
`Mill’s first method is the Method of Agreement: if A is always followed by a,
`then A is presumably the cause of 4. Mere agreement does not, however, furnish
`rigorous proof, although you may be limited to it when you lack the voluntary
`variation of events—the independent experimental variable—and are reduced to
`description only. For this reason the establishment of causal relations in biography,
`history, geology, paleontology, and even astronomy is less sure than in experimental
`science. Mill remarked that mere agreement would indicate that night is the cause of
`day, and day the cause of night, since the sequence is universal, and he noted that we
`can be more certain that agreement indicates cause when the antecedent term in the
`conjunction of events can be established at will without dependence upon other
`events. Mill was right in mistrusting the Method of Agreement, since the concurrence
`of A and 4 in sequence may mean only that both are elfects of the same sufficient
`cause, and, since if that other cause is suflicient but not necessary, it takes the Method
`of Difference to show that A and a are not necessary concomitants. It is for this
`reason that Mill suggested that the Method of Agreement is strengthened if A can
`be varied “at will,” that is to say, if A is a freely independent variable. Such a caveat,
`however, actually constitutes an extension of the Method of Agreement to include the
`Method of Difference (when variation of A includes its elimination) or the Method
`of Concomitant Variation (when A is merely changed in degree). The inference of
`causation is never safe when based upon agreement alone.
`The Method of Diflerence is Mill’s second method: if A is always followed by a,
`and not-A is always followed by fl0t‘d,
`then A is certainly the cause of a. This is
`equivalent to adding the control observation:
`if not-A,
`then not-4. Mill used the
`word control once: "It thus appears that in the study of various kinds of phenomena
`which we can, by our voluntary agency, modify or control, we can in general satisfy
`the requisitions of the Method of Difference; but that by the spontaneous operations
`of nature those requisitions are seldom fulfilled." This use of control is, however, in
`
`controlling environment; Self-control; Personal control; Group control; Economic
`control; Culture and control; and The problem of control. Certainly Solomon's
`control group has little relation to Slcinner’s group control. Even in discussion of
`experimental design the meaning of control is apt to vary. For instance, C. E. Buxton,
`in T. G. Andrews, Method: of Psychology, 1948, 73 f., in the course of two pages
`uses the word thrice as meaning maintenance of constancy of conditions and once in
`the sense of a test-observation.
`5 J. S. Mill, A System of Logic, Ratiocinative and Inductive, 1843, Bk. III, chap. 8.
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`3
`
`
`
`NATURE AND HISTORY or EXPERIMENTAL CONTROL
`
`575
`
`the sense of direction or guidance, whereas the Method of Difference provides control
`in the sense of a verifying check, although Mill did not make that use of the word.
`He recognized, however, the fundamental relation of his first two methods, speaking
`of the Ioint Method of Agreement and Diference, which is essentially the modern
`scientific procedure for treating contingencies when continuities are not observed.
`(The third method, the Method of Reriduer, need not concern us. If ABC is known
`to be the cause of abc, and BC the cause of bc, then A must be the cause of a, even
`though A can not be produced without BC nor thus a without 17:.)
`Mill's fourth procedure is the Method of Concomitant Variations. Nowadays we
`think of such observation as basic to all experiments and thus of Agreement and
`Diflerence as special cases of Concomitant Variation. Concomitant variation exists
`when there is a series of diflerences, and in any pair of concomitances one concomi-
`tance furnishes a comparison or control for the other. So we could get along with
`this method alone, if it were broadly enough conceived, except for the historical fact
`that the concept of control actually grew out of the consideration of the Method of
`Difference, which Mill's prestige established as independently important.
`Mill is usually taken as the authority on this matter. Thought about these prin-
`ciples is, however, historically continuous. A century earlier Hume had a similar, less
`specific discussion, which laid down rules equivalent to the Method of Difference
`and to the Joint Method of Agreement and Difference.“ Still another century earlier
`Francis Bacon discussed the collection of data by the finding of instances that agree,
`and of negative instances and cases to furnish comparison.’ In these authors the
`anticipation of Mill is quite clear if one but remembers what a century or two can
`do to both thought and its expression.
`At this point it is important to remark that every statement of fact expresses some
`kind of a difference. Even such description as is not experimental, being specific and
`thus discriminative, diflerentiates what
`is from what is not. Jevons oflered as the
`formula for a fact: “Where A is, X is; and where A is not, X is not;”‘ and that, of
`course, is also a statement of agreement and difference. Jevons remarked: "Every cor-
`rect and conclusive experiment necessarily consists in the comparison of results between
`two different combinations of circumstances.” It is plain then that i-n scientific descrip-
`tion we are not going to get away from the concept of control, although the idea
`appears with different degrees of specificity and formality in statements of experi-
`mental design. If you have an observed datum, there is always some point or frame of
`reference in respect of which the datum makes sense, and it might aid clarity of
`thought if one were -to think always of a datum set over against a relatum. (‘If A, then
`a) would be a datum, and (if not-A,
`then not-a) would be the relatum and the
`control observation, which might indeed depend upon a control-experiment to establish
`it. Again and again when agreement seems to yield certainty, it is because the control-
`observation is implied or even included in the experiment.
`Now let us examine some instances of the use of the concept of control and of the
`word control in the scientific literature of the last one hundred years.
`
`‘ David Hume, A Treatise on Human Nature, 1739, Bk. I, pt. iii, sect. 15.
`' Francis Bacon, Nor/em Organurn, 1620, Bk. II, Aph. 11-13, which deal respec-
`tively with the three principles.
`3 W. S. Jevons, Principle: of Science, '1st ed., 1874, II, 44; or 2nd ed., 1883, 433.
`’_Ievons, op. cit., 1874, II, 43; or 1883, 433.
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`4
`
`
`
`576
`
`BORING
`
`Control as restraint or guidance. The meaning of control as restraint or
`guidance is the common, though later, meaning of the term, and in science
`it applies to keeping experimental conditions constant and also to alter-
`ing the independent variable in accordance with precise known prede-
`termination. We do not need to hunt out instances of the scientific aspira-
`tion to keep conditions constant,
`to maintain controlled conditions or to
`vary a parameter under controlled conditions. In these contexts control,
`constancy and precision are of the essence of experimental science. Thus
`Titchener, in describing the nature of experiment, emphasized the neces-
`sity for repetition,
`isolation, and variation in the interests of good ob-
`servation, constancy of conditions and exact variation, although he did
`not use the word control.” In similar sense Murphy and Murphy have
`spoken of
`the “relatively uncontrolled observational and biographical
`studies" of children and the later "new experimental and highly controlled
`observation studies worked out" after 1915.11 We can also go back to
`Fechner who sought constancy of conditions by control of experimental
`procedures and the treatment of data: “The arrangement of experimental
`conditions,” he wrote, “the recording of observed values, the enumeration
`of errors or of right and wrong cases, as well as all the calculations based
`upon them, must be so arranged and controlled (controliren) by repetition
`and otherwise that, as far as possible, errors are avoided by the multiplica-
`tion of data, by calculation when error is otherwise unavoidable, and by
`observing an immutable integrity in the recording and conversion of
`data."12
`
`implies
`independent variable, moreover,
`Control of the experimental
`guidance as well as the maintenance of constancy, but the same admonitions
`and aspirations apply to this kind of control as to the control of conditions
`fixed throughout a particular experiment.
`Control enters into psychic research with two different meanings. Holding hands
`and touching feet with each of your neighbors in the spiritistic circle may be thought
`to promote psychic continuity, but the practice also acts as a control against fraud.
`There is also, however,
`the spirit control, who or which guides and directs the
`
`1'’ E. B. Titchener, A Text-Book of Psyclaology, 1910, 20; A Beginner's Psychology,
`1915, 22-25. The only use of the word control that I can find in these books is in
`Text-Book, 22, where Titchener remarks that “experimental control is still possible"
`even when introspection interferes with consciousness.
`“ Gardner Murphy and L. B. Murphy, Experimental Social Psychology, 1951,
`201 f., with examples of systematic arrangements for the recording of exact behavior
`and the maintenance of constancy of conditions on pp. 214-227.
`" G. T. Fechner, Elernente der Psytbop/Jysié, I, 1860, 85; and there is another
`example of the use of the word farther down on the same page: "die Wiederholung
`oder sonstige Controle an sich langweiliger Operationen."
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`5
`
`
`
`NATURE AND HISTORY or EXPERIMENTAL CONTROL
`
`577
`
`medium. This use of the term seems to have come in late, for Podmore makes no
`mention of it in 1902 in his excellent and thorough two-volume history of spiritual-
`ism.“ Nor did the New Englirh Dictionary give this meaning in 1893, although the
`Standard Dictionary by 1913 was saying for one meaning of control: “The intelli-
`gence (whatever its nature) which regulates the communication of messages through
`a medium or psychic."
`
`Control obrernationr, Jerier, and experiments. We come back to the
`special use of the word control in its original meaning as a check or
`standard of comparison, a relatum. If separate observations have controls
`interspersed (e.g. single points introduced among the double points which
`are used in determining the two-point limen upon the skin), then we have
`what may be called control obrerz/ationr. When the controls are or-
`ganized into series, then we have a control rerie: with which the experi-
`mental rerier is compared, as in various memory experiments. If the
`dissociation of the terms for comparison is greater and the organization of
`each more elaborate, then we begin to speak of the principal experiment
`and the control experiment. The term control experiment, as a standard
`of comparison, has now got into most of the dictionaries. Its synonym is
`text experiment. Logically there is, however, no difference between one
`kind of relatum and another except in degree of organization and formal
`independence. The basic conception, of course, appears early in experi-
`mentation, whereas the use of the word control comes later.
`
`The concept of control is pretty old and was quite obvious once the Renaissance
`had turned men's thought from theological fiat to experiment as the means for pene-
`trating into nature's secrets. Here is a story that makes the whole matter clear.
`In 1648 the Torricellian vacuum was known to physics in general and to Pascal
`in particular. This is the vacuum formed at the upper closed end of a tube which has
`first been filled with mercury and then inverted with its lower open end in a dish of
`mercury. The column of mercury falls in the tube until it is about 30 in. high and
`remains there, leaving a vacuum above it. Pascal was of the opinion that the column
`is supported by the weight of the air that presses upon the mercury in the dish (he
`was right;
`the Torricellian tube is a barometer) and that
`the column should be
`shorter at higher altitudes where the weight of the atmosphere would be less. So
`he asked his brother-in-law, Perier, who was at Clermont, to perform for him the
`obvious experiment at
`the Puy-de-Dome, a mountain in the neighborhood about
`3000 ft. (“S00 fathoms”) high as measured from the Convent at the bottom to the
`mountain's top. On Saturday, September 19th, 1648, Perier, with three friends of
`the Clermont clergy and three laymen, two Torricellian tubes, two dishes and plenty
`of mercury, set out for the Puy-de-Dome. At the foot they stopped at the Convent,
`set up both tubes, found the height of the column in each to be 26 old French
`inches Plus 31/2 Paris lines (28.04 modern inches),
`left one tube set up at the
`
`“ Frank Podmore, Modern Spiritualirm, Hirtory and Criticimz, 2 vols., 1902.
`
`This content downloaded from 2l6.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`6
`
`
`
`578
`
`BORING
`
`Convent with Father Chastin to watch it so as to see whether it changed during the
`day, disassembled the other tube and carried it to the top of the mountain, 3000 ft.
`above the Convent and 4800 ft. above sea-level. There they set
`it up again and
`found to their excited pleasure that the height of the mercury column was only 23
`French inches and 2 Paris lines (24.71 in.), much less than it was down below just
`as Pascal had hoped it would be. To make sure they took measurements in five
`places at the top, on one side and the other of the mountain top, inside a shelter and
`outside, but the column heights were all the same. Then they came down, stopping on
`the way to take a measurement at an intermediate altitude, where the mercury column
`proved to be of intermediate height (26.65 in.). Back at the Convent, Father Chastin
`said that the other tube had not varied during the day, and then, setting up their
`second tube, the climbers found that it too again measured 26 in. 31/2 lines. These
`are reasonable determinations for these altitudes, showing about the usual one inch
`of change in the mercury column for every 1000 ft. of change in altitude.“
`In this experiment there was no elaborate design, and it took place 195 years too
`soon for the experimenters to have read John Stuart Mill's Logic, but the principle
`of control and of the Method of Difference is there. How important it was for them
`to have left a barometer at the base of the Puy-de-Dome to make sure that changes in
`the tube that they carried up the mountain were due to elevation and not to general
`atmospheric changes or to other unknown circumstances! How wise of the party at
`the top to have made the measurement under as many different conditions as they
`could think of with altitude constant! How intelligent of them to take a reading on
`the way down and thus to turn the Method of Difference into the Method of Con-
`comitant Variation!
`
`When Jevons was writing his section on “Blind or Test Experiments,” he missed
`this paradigm from Pascal (1648) but gave one from Faraday (1848) and another
`from Tyndall (1865).” Neither of these examples is in any formal sense a blind
`experiment or, as we should say today, a control experiment. They are simply ex-
`amples of the basic principle that a fact is a relationship and that you are not ready
`to make a scientific statement until you have a comparison to present. Thus Faraday,
`in the research cited by Jevons as including blind experiments, was really trying out
`what he called magnecrystallic action on a variety of difl'erent substances, finding
`that certain rules of magnetic orientation hold for crystals of bismuth and similar
`metals and that there is no magnetic action of this kind at all for the crystal-s of
`certain other metals, of which lead is one example." Faraday was engaged in what
`
`"‘ Blaise Pascal, The Phyricnl Treatirer of Parcel: the Equilibrium of Liquid: and
`the Weight of the Mars of the Air, trans. 1937, 103-108, which is mostly Perier's
`letter to Pascal about the experiment. For short accounts, see I. B. Cohen, Science,
`Servant of Man, 1948, 71 f.; J. B. Conant, Science and Common Sense, 1951, 72-74.
`Cohen says that Perier noted the continuous drop in the height of the column as he
`came down; if three points make a continuity, he did. Conant says that Pascal was at
`the foot of the mountain watching the control barometer, but that was Father Chastin.
`Pascal was waiting in Paris to hear what happened.
`“ Jevons, op. cit. (note 8), 1 ed., 1874, II, 45-45, is the section called “Blind or
`test experiments," in the chapter on “Experiment”; but the mention of Faraday’s and
`Tyndall's experiments does not appear until 2 ed., 1883, 433 f.
`“‘ Michael Faraday, Experimental Rerearche: in Electricity, 1855, III, Series xxii,
`83-136. The experiments on magnecrystallic action were performed in 1848.
`
`This content downloaded fi'0m 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`7
`
`
`
`NATURE AND HISTORY OF EXPERIMENTAL CONTROL
`
`579
`
`might be called the experimental taxonomy of metallic behavior, and, now that his
`results are available to us, we can separate them into data and relata, into experi-
`ments and controls,
`though Faraday did not see his problem thus. Tyndall, in the
`place cited by Jevons, was describing an experiment in which 19 lb. of water were
`brought from room temperature to a boil in 21/; hr. by the constant rotation of a
`steel cylinder fitting tightly inside another steel cylinder and kept by a horse in rota-
`tion while immersed in the water—a demonstration that mechanical power can be
`turned into heat by friction." Presumably Jevons believed that the control here was
`the initial condition of the water at room temperature, but actually this experiment
`was an example of the Method of Concomitant Variation, for the temperature of the
`water was taken from time to time and found steadily to rise as equine energy was
`continuously converted into heat.
`
`In 1870 a new kind of control observation appeared in psychological
`experimentation,
`the Vexirverruc/9 that Vierordt—and later Riecker in
`1874——used in their determinations of the cutaneous two-point
`thresh-
`old.
`
`If you keep putting two points simultaneously down upon the skin with varying
`separations between them and keep track of the number of times each separation is
`perceived as a single impression and the number of times it is perceived double,
`then, if your chosen separations are proper, you can compute the separation at which
`a two would be felt as often as a one. That is the threshold. The observer, well trained
`to give phenomenal reports, can participate in this experiment quite successfully, but
`the naive observer is apt to make what Titchener called the stimulus-error, to report
`two often or always because he knows or guesses that the stimulator often or always
`puts two points to the skin. So Vierordt and then Riecker introduced single stimuli
`as checks upon the observer, calling them Vexirr/ermc/ae or puzzle trials-—actually
`control observations. Riecker worked with the method of constant stimuli which was
`
`still called then, after ‘Fechner’s original title, “the method of right and wrong cases.”
`It is ‘right’ to call two points two and one point one, and ‘wrong’ to call two points
`one and one point two. Actually the observers often did call one point two for there
`is a special physiological disposition for getting a two-fold impression out of a
`unitary stimulus, and these errors with the Vexiri/ermcbe were called Vexirfebler
`(paradoxial errors)?’
`that there should be as
`Later McDougall formally introduced the requirement
`many single stimuli as double and that the threshold should be 80% right for both
`
`" John Tyndall, Heat Considered a: 4 Mode of Motion, 1863, 21.
`” Karl von Vierordt, Die ‘Abhingigkeit der Ausbildung des Raumsinnes der Haut
`von Beweglichkeit der Korperteile, Zrcb. Bt'ol., 6, 1870, 53-72; Adolph Riecker,
`Versuche fiber den Raumsinn der Kopfhaut, ibid., 10, 1874, 177-201. On this experi-
`ment, the Vexin/er.mc17e and the Vexirfe/aler, see Victor Henri, Ueber die Raum-
`wa/arnelmzungen der Tartsinner, 1898, 58-89, esp. 61-66. On the nature of Vexir-
`felzler, which are no more "errors” than is any compulsive illusion, see Margaret
`Kincaid, An analysis of the psychometric function for the two-point limen with respect
`to the paradoxical error, this JOURNAL, 29, 1918, 227-232; Harry Helson and R. M.
`Burgert, Prediction and control of judgments from tactual single-point stimulation,
`ibid., 48, 1936, 609-616.
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`8
`
`
`
`580
`
`BORING
`
`kinds of stimuli taken together." That was good functionalism at the time and is
`nowadays good modern behaviorism: you see how adequate the subject
`is to his
`environment
`(stimulus~world) when without further knowledge he relies for in-
`formation upon the sensitivity of his receptor apparatus, but Titchener still thought
`that a trained observer should be able to judge sensory impression accurately in its
`own right, Whatever he suspected of the stimulus.” McDougall stuck by his guns
`even for educated subjects, and Henry Head and his associates used McDougall's
`method in the experiments which set up the distinction between protopathic and
`epicritic sensibility.” It
`is only recently that
`the Vexirverxuclze have come to be
`regarded as control observations, and indeed they are somewhat special for they func-
`tion as straight checks or controls for objective behavior, whereas they may also
`become directing guides in an introspective psychology where a trained observer
`needs to be checked as to whether he is reporting only on sensory impression or
`whether, committing the stimulus-error, he is trying to infer or guess the true nature
`of the stimulus-object."
`
`It was along in the 1870s that the word control began to be used in the
`sense of a check or a standard of comparison in respect of which a differ-
`ence is expected to lie.
`
`In 1874, in the first edition of his famous P/ayriologisc/ye Prycbologie, Wundt said
`that the fall-apparatus, used then to calibrate the Hipp chronoscope,
`is a control.
`Later the fall-apparatus changed its form and became and was called a control Imm-
`mer. Proper procedure is to run a series in which the times of the successive falls
`of the control hammer are measured by the Hipp chronoscope, and that series, which
`Wundt actually called a Korztrollverruc/2, gives you the variable error of the chrono-
`scope and, if you have measured the true time of fall on the chronograph, the constant
`error too.”
`Thus Wundt was ready enough to control a piece of apparatus with an objective
`check, but the only control he admitted as valid for the human observer was rigorous
`training in psychological observation. For psychological experimentation he registered
`objection to Korttrolloermcbe, Prfifungwerrucbe, Probeverwc/re and Vexirvermc/ye."
`
`“William McDougall, Cutaneous sensation, Report: of the Cambridge Anthro-
`pological Expedition to Torre: Strain, 1905, II, pt. ii, 189-195.
`“Titchener, On ethnological tests of sensation and perception, with special refer-
`ence to tests of color vision and tactile discrimination in the reports of the Cambridge
`Anthropological Expedition to Torres Straits, Proc. Amer. Pltilos. Soc., 55, 1916,
`204-236, esp. 206-215.
`2‘ Henry I-lead, Studies in Neurology, 1920, I, 26-29. The original experiment was
`Head, W. H. R. Rivers and James Sherren, The afferent nervous system from a new
`aspect, Brain, 28, 1905, 99-115 (reprinted in op. cit., 1920, 55-65); Head and Rivers,
`A human experiment in nerve division, Brain, 31, 1908, 325-450 (reprinted in op.
`L'it., 1920, 225-329).
`” Cf. E. G. Boring, The stimulus-error, this JOURNAL, 32, 1921, 449-471, which,
`paradoxically,
`interprets Titchener’s conception of the psychological point of view
`functionally.
`”Wilhelm Wundt, Grundziige der pbysiologisclyen Pryclaologie, 1st ed., 1874,
`772; 6th ed., III, 1911, 367. See also Titchener, Experimental Pryclaology, II, 1905,
`pt. ii, 340-344.
`“Wundt, op. cit., 6th ed., III, 1911, 399.
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`9
`
`
`
`NATURE AND HISTORY or EXPERIMENTAL CONTROL
`
`581
`
`Also in the 1870s this meaning of the word control entered biology by way of
`Darwin. In 1875 he reported on Utricularia, plants that float on foul water and that
`carry tiny bladders with valves that
`trap the insects and organic matter that get
`within the orifices of the bladders. It is a study in the taxonomy of plant behavior.
`After observing the reaction of the bladders to their normal food, “four bladders
`were tried as a control experiment," Darwin wrote. He fed them gum arabic and
`also sugar, neither of which produced the normal feeding reaction, whereas nitrate
`of ammonia did.” Later Darwin is found describing the movement of plants in
`response to irritants. He performed experiments with the radicles of peas, attaching
`to them as irritants bits of cardboard “which served as standards of comparison or
`controls."” In 1890 there was an article in Nature on the immunization of mice
`against tetanus. Immunized mice were inoculated with tetanus and failed to show its
`symptoms, but "control mice [not immunized] died within 48 hours.""’ When a
`word can be used thus casually, it may be assumed to have become established in the
`language-—-in this case in the vocabulary of biology.
`
`As experimental psychology developed, the design of its experiments be-
`came more rigorous and elaborate. Speaking approximately, we may say
`that formal design first developed in psychophysics, then in the reaction
`experiments, then in the experiments on memory.
`
`In 1900 we find Miiller and Pilzecker in their classical study of right associates
`using comparison series (Vergleirlareiben) as controls (but not so named) for the
`principal series (Hauptreilaerz). They also introduced fore-series (Vorreiberz) and
`after-series (Nacbreilaerz) and thus in a way they anticipated some of the design
`that Thorndike and Woodworth employed shortly after for their study of transfer.
`You even find in Muller and Pilzecker Hrmpzr/orreilaerz, Vergleiclworreiberz, H4145!-
`mzrbreilaerz and Vergleiclmacbrei/aerz, but the Vergleicbreilaen are control series used
`on the same observers. The control groups come later.”
`
`There can be no doubt that use of control observation, either implicit
`or explicit,
`is essential in sound experimental work. There has to be a
`relatum to give the datum significance. More definitely designed control
`series and control experiments have at the present time come into common
`use, even when control groups are not called for. Such controls go by a
`variety of names. For instance,
`in medical research one sometimes sees
`nowadays mention of the use of placebo (a dose that pleases the patient
`but has no pharmacological effect; placere,
`to please)
`introduced as a
`control in comparison observations or in control series.”
`
`" Charles Darwin, lrzrectivoraur Plarztr, 1875, chap. xvii, esp. 413.
`"" Charles Darwin and Francis Darwin, Power ofMa1/emerzt in Plants, 1881, chap.
`iii, esp. 160-163.
`2' E. H. ‘I-Iankin, A cure for tetanus and diphtheria, Nature, 45, 1890, 121-123.
`”G. E. Miiller and Alfons Pilzecker, Experimentelle Beitréige zur Lehre vom
`Gediichtniss, Zrclr. Pryrbal., Ergbd. 1, 1900.
`“Stewart Wolf, Effects of suggestion and conditioning on the action of chemical
`
`This content downloaded from 216.43.235.227 on Mon, 13 Jun 2016 13: 16:32 UTC
`All use subject to http://about.jstor.org/terms
`
`10
`
`10
`
`
`
`5 8 2
`
`BORING
`
`When there is no acceptable hypothesis as to the independent variable
`in an experiment, it may be impossible to have a control, which may become
`available only when a good hypothesis comes along. Thus, in the experi-
`ments on extrasensory perception (ESP), control is very difficult. Control
`by restraint and guidance is common enough for every effort is made to
`keep conditions constant and to prevent fortuitous leakage of information
`to the percipient, but control as a check requires a knowledge of what is
`to be omitted if the Method of Difference is to be used. The reading of
`cards in an invisible pack is supposed to have been extrasensory when the
`reading deviates from ‘chance’ by an amount that can be accepted under
`current statistical conventions with a high level of confidence. But what is
`‘chance’? Good statisticians nowadays do not accept, as they did thirty years
`ago, the principle of insufficient reason as meaning the a priori presumption
`that wholly unconstrained coincidence will give correct guesses in only