steps and pips in the history of the cumulative recorder · key words: cumulative record,...
Embed Size (px)
JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR 2004, 82, 329–355 NUMBER 3 (NOVEMBER)
STEPS AND PIPS IN THE HISTORY OF THE CUMULATIVE RECORDER
KENNON A. LATTAL
WEST VIRGINIA UNIVERSITY
From its inception in the 1930s until very recent times, the cumulative recorder was the most widelyused measurement instrument in the experimental analysis of behavior. It was an essential instrumentin the discovery and analysis of schedules of reinforcement, providing the first real-time analysis ofoperant response rates and patterns. This review traces the evolution of the cumulative recorderfrom Skinner’s early modified kymographs through various models developed by Skinner and hiscolleagues to its perfection in the 1950s, and then into the 1960s when it proliferated as differentscientific instrument companies began marketing their own models of the cumulative recorder. Withthe rise of digital computers, the demise of the cumulative recorder as a scientific instrument wasinevitable; however, the value of the cumulative record as a monitoring device to assess schedulecontrol of behavior continues. The cumulative recorder remains, along with the operant condition-ing chamber, an icon of Skinner’s approach to psychology.
Key words: cumulative record, cumulative recorder, kymograph, rate of response, F. S. Keller, B. F.Skinner
Cumulative records were the product of cu-mulative recorders, instruments used widelyduring most of the history of the experimen-tal analysis of behavior. The cumulative re-corder was to early behavior analysts what themicroscope was to early biologists: a windowonto a world difficult, if not impossible, to
Many people contributed valuable information thatfound its way into this review, either directly or indirectly,but none more than Mike Harrison of Boston University,who also provided a Harvard Instruments Company ky-mograph and a set of Ralph Gerbrands Company record-ers, Models C-1 through C-3, to a museum dedicated topreserving behavioral research instruments at West Vir-ginia University. Jim Dinsmoor and Bill Morse generous-ly, and patiently, taught me about the early cumulativerecorders at Indiana and Harvard, respectively. Tim Boonof the British Museum of Science in London most gen-erously gave me access to the Museum’s extensive ky-mograph, psychology, and instrumentation catalogue col-lections. A J. R. Kantor fellowship from the Archives ofthe History of American Psychology in Akron, Ohiomade it possible for me to glean valuable backgroundinformation from the Archive’s psychology instrumentcollection and from several delightful and informative in-teractions with John Popplestone and Marian White Mc-Pherson. I also thank Peggy Kidwell of the SmithsonianMuseum in Washington, D.C. for arranging access to theSkinner bequeathals in the Smithsonian collection. Agrateful pip of the pen as well for technical assistanceand information about details of many of the early cu-mulative recorders and manufacturers to Cindy Ander-son, Roger Bunce, Jim Chambers, Dave Eckerman, JoelGreenspoon, Tatu Hunziger, Frances Keller, John Keller,Vic Laties, Lew Gollub, Bruce Overmier, Mike Perone,Phil Reed, Murray Sidman, Warren Street, Walter Sullins,Julie Vargas, and Toshihiko Yoshino.
Address correspondence to K. A. Lattal, Departmentof Psychology, West Virginia University, Morgantown,West Virginia 26506-6040 (e-mail: [email protected]).
study by unaided observation. A cumulativerecorder typical of those manufactured in thelate 20th century (Ralph Gerbrands Compa-ny Model C-4) is shown in Figure 1. Housedin a metal chassis, a continuous roll of paperwas fed across an aluminum platen at a con-stant speed (typically about 31 cm, or 1 ft, perhr, but this often could be adjusted by achange of gears or switch setting), marked byone or more pens to show behavioral changesreflecting the contingencies of consequencesin effect. In the usual experimental arrange-ment the recorder had four functions: (a) aresponse pen stepped vertically across the pa-per with each response; (b) if so pro-grammed, the response pen pipped (i.e., de-flected) typically downward with eachreinforcer; (c) the response pen holder resetto the bottom of the paper when the penholder reached the top of the page (typicallyafter 550 responses, but this too was adjust-able as well with many recorders); and (d)one or more separate event pens deflected upor down to indicate the occurrence of othersignificant environmental events, such as se-lected responses or stimulus changes.
Thus the device created a cumulativegraph of operant responses, along with theconcurrent indication of other significantevents. These graphs came to be known ascumulative records (see Morris & Smith for a discussion of the history of theterm), examples of which, generated by ex-posure to different schedules of reinforce-
330 KENNON A. LATTAL
Fig. 1. A Ralph Gerbrands Company Model C-4 cu-mulative recorder showing (A) stepping pen; (B) eventpen; (C) platen and recording paper; stepping pen hold-er forward and to the left of the letter (D), microswitchfor initiating reset of the recorder immediately behindthe letter (D); (E) manual reset switch. Each reset of thestepping or response pen (line marked F) represents thecompletion of 550 responses, and each downward deflec-tion (pip) of the response pen indicates a reinforcer de-livery.
ment, are shown in Figure 2. Cumulative re-cords gave the investigator visual access tobehavior in real time like no other measurebefore, revealing both small, local, behavioraleffects and more molar ones such as the char-acteristic temporal and response patternsgenerated by each type of reinforcementschedule (for a comprehensive review of cu-mulative records generated by differentschedules of reinforcement, see Ferster &Skinner ).
The history of the cumulative recorder isthe story of gaining control over the fouraforementioned functions: step, pip, reset,and event mark. Thus it is a history of strivingto achieve an ever more accurate and precisepicture of behavior in real time, the primarysubject matter of the discipline. In the broad-er scheme of things, it is also in microcosmthe story of the experimental analysis of be-havior and how the reciprocal interaction be-tween the scientist, the subject matter, and itsmeasurement has led to change and pro-gress.
The story may be divided into several parts.It begins with the invention of the kymo-graph, long before Skinner comes on to thestage of American psychology. The story thentraces the cumulative recorder through itsbirth in Skinner’s hands, its refinement by
Skinner and his colleagues, its proliferationas a commercial product manufactured byscientific instrument companies, and its ex-tinction as both the science and its technol-ogy outstripped the need for cumulative re-cords as primary data.
Skinner’s Wonderful Machines
B. F. Skinner was, by all accounts (e.g.,Bjork, 1993; Ferster, 1970, p. 39; Morse &Dews, 2002, p. 314; Skinner, 1976a), a tinker-er. He had a mechanical bent and was alwaysmaking things. His career was filled with thesuccessful outcomes of such tinkering: crea-tive constructions of various pieces of appa-ratus for conducting operant conditioningexperiments on rats and pigeons, for teach-ing students of psychology about his concep-tual framework, for maintaining a comfort-able environment for his daughter, and formaking his work life more efficient and fo-cused (e.g., Trudeau, 1990). At least two ofthe outcomes of his tinkering have achievediconic status in psychology: the operant con-ditioning chamber (a.k.a. the Skinner box, alabel he disliked; see Lattal, 1999) and thecumulative recorder. There is a reciprocal in-teraction between these two pieces of appa-ratus and the essential features of his scienceof behavior. One is nearly impossible to sep-arate from the other, at least in the early daysof the science. In a small masterpiece of theart of inductive science, Skinner (1956) pro-vided a history of the evolution of the oper-ant conditioning chamber. He described hisearliest attempts to depict the behavior of or-ganisms in real time, but this story ends onlyat the beginnings of cumulative recording. Totrace the remainder of that history, we mustfirst go back to a time before the events ofSkinner’s case history, for only then will wehave a context for the history of cumulativerecorders that followed.
Why Cumulative Recording?
But first, the question of why depict behav-ior at all must be addressed. For Skinner, be-havior was a subject matter of interest in itsown right and ‘‘an emphasis on rate of oc-currence of repeated instances of an operantdistinguishes the experimental analysis of be-havior from kinds of psychology which pro-
331HISTORY OF THE CUMULATIVE RECORDER
Fig. 2. Top: Examples of cumulative records from complete experimental sessions generated by a Ralph GerbrandsCompany Model C-2 cumulative recorder. Responses of 2 rats, 46 and 53, were maintained on a mixed variable-interval (VI) 1-min variable-time (VT) 1-min schedule of reinforcement, where single 45-mg food pellets served asthe reinforcers. The mixed-schedule component duration was 10 min in the top session and 30 min in the bottomone. The response pen reset with each component change. The event pen was up during the variable-interval com-ponent and down during the variable-time component. Reinforcers are marked by downward deflections (pips) ofthe response pen (excerpted from Lattal, 1973). Bottom: Records of pigeons’ key pecking maintained on a fixed-ratio 110 (left) or a variable-ratio 110 (right) schedule with access to mixed grain as the reinforcer. Presumably therecords were generated by early model Ralph Gerbrands Company cumulative recorders, but the details of the modelsare unknown. As in the top set of records, pips indicate reinforcers, but in this early version the pip was generateddifferently and appears as a hash across the record rather than as a downward deflection of the pen. The recordshave been telescoped to make possible the presentation of more data in less space, a procedure described in detailby Ferster and Skinner (1957), from which these records were taken (excerpted from Figures 96 and 479 of thatvolume).
ceed in . . . [other] ways’’ (1966, p. 213).‘‘Rate of responding is a basic dimension, notsimply because responses can be accuratelycounted, but because rate is relevant to thecentral concern of a science of behavior’’(1966, p. 214). Skinner also was interested inmoment-to-moment changes in behavior inreal time as a function of environmentalchanges; for example, the addition or remov-al of reinforcers or other stimuli:
An emphasis on real time is another reasonwhy cumulative records are useful. (A cumu-lative record is sometimes used to ‘‘smooth’’other kinds of data—for example, the errorsmade during repeated trials in learning amaze or in solving a problem—and it is oftenimplied that a cumulative record of responsesin time also gains an unwarranted smoothnessof the same sort. The important difference isthat the slope of a cumulative curve in real
332 KENNON A. LATTAL
time represents a meaningful state of behav-ior.) (1966, p. 216)
Ferster (1953) observed that:
Whenever an intermittent schedule of rein-forcement is used, a cumulative record gen-erally proves to be the most convenient anduseful method of recording. . . . The virtue ofthe cumulative record is not that it allows aprecise measurement of the rate at any partic-ular time, but rather that it emphasizes chang-es in rate which can be seen in the curvaturesof the record. (p. 267)
More broadly, in contrast to his own posi-tion, Skinner asserted that most theories oflearning posited that ‘‘the learning goes onsomewhere else, in a different dimensionalsystem’’ (1950/1972, p. 72). The cumulativerecorder made learning a truly empiricalmatter, something that could be directly ob-served as ongoing responding was traced outin a graph. Estes (personal communication,December 21, 1996) commented on another,more practical, effect with respect to one ofSkinner’s early recorders (Heron & Skinner,1939): ‘‘It made a good show for visitors, whocould, in Skinner’s manner of speaking, seebehavior being emitted.’’ Parsonson and Baer(1986) summarized simply a general ratio-nale for the graphic analysis of data such asthat found with cumulative records: ‘‘Thegraph serve(s) as a comprehensive yet simplemeans of recording, storing, representing,communicating, and, above all, analyzing be-havioral data’’ (p. 157).
Morse and Dews (2002), however, madethe most perceptive statement about the cu-mulative recorder’s unique place in the his-tory of the science of behavior: ‘‘It was thecumulative recorder that permitted the rec-ognition of the powerful effects of schedules[of reinforcement]. . . . The characteristicproperties of different schedules would nothave been discovered without the cumulativerecorder’’ (pp 313–314). The cumulative re-corder’s importance in leading to an under-standing of how subtle, and not so subtle,changes in the environment are functionallyrelated to subtle, and not so subtle, changesin behavior cannot be overstated. Simply put,without the cumulative recorder to generaterecords of the behavior of interest, therewould have been no experimental analysis ofbehavior. At a more individual level, the cu-
mulative recorder allowed the powerful im-mediate consequence for the scientist of see-ing the effect of experimental manipulationsalmost as they were made. There was no needto wait for a summary graph or a statisticalanalysis to determine whether an effect wasto be had, the effect was a matter of (cumu-lative) record (see Michael  for a fur-ther discussion of the role of immediacy ofresults in determining scientific behavior).Such a consequence likely was a factor indrawing several generations of behavior ana-lysts to the discipline.
At the time of Skinner’s early work, the re-cording of behavior typically was done eithermanually or mechanically, the latter for themost part graphically by means of an instru-ment adapted from experimental physiology,the kymograph. Digital counters were notgenerally used in experimental psychologyuntil some time after Skinner’s earliest ex-periments. Thus it is not surprising that theearliest recording of operant responses wasby means of a kymograph, as described bySkinner (1956).
The kymograph was invented by the Ger-man physiologist Karl Ludwig (1816–1895) in1847 and was first used by him to study bloodpressure in nonhuman animals. An exampleof a kymograph typical of those used duringthe late 19th and early 20th centuries isshown in Figure 3. A sensor attached to a lo-cation of interest on (or in) the organism wasin turn attached to a transducer. The trans-ducer then was attached to a stylus in contactwith a piece of paper wrapped around a met-al drum. The paper was smoked with carbonblack (before ink and pens were used, atleast) so whenever the drum moved the stylusscraped away a bit of the carbon on the blackpaper, leaving a white mark. So, through me-chanical linkages the stylus was deflected byphysiological events and because the drumrotated at a constant speed (being driven bya motor), a recording of these events in realtime was achieved. (At least into the 1960s,Skinner gave to his graduating Ph.D. studentsautographed copies of a cumulative recordscratched on a smoked piece of kymographpaper [preserved by a varnish-like fixer] fromhis early experimental work [see Figure 7 inCatania, 2002].)
333HISTORY OF THE CUMULATIVE RECORDER
Fig. 3. Drawing of an early kymograph, with carbon-blackened paper being marked by a stylus.
Fig. 4. A Harvard Instruments Company kymographlike that used by Skinner in his early attempts to createcumulative records.
The kymograph, or Ludwig’s drum, rapidlybecame a mainstay of physiological laborato-ries around the world, undergoing some, butnot extensive, modifications and refinementsas time passed. By the time Skinner went towork in W. J. Crozier’s physiology laboratoryat Harvard in 1928, kymographs were un-doubtedly about the laboratory, making theiradaptation to his psychological experimenta-tion highly probable. In fact, the kymographhad been used earlier by psychologists andothers who were interested in recording in-dividual responses in real time. For example,in the first American psychology textbookJames (1890) presented a kymograph record(or kymogram) to illustrate reaction time.The famous diagram of Pavlov’s preparationfor classical conditioning shown by Yerkesand Morgulis (1909, their Figure 2) promi-nently depicts a kymograph as part of the ap-paratus. And the British psychologist Grind-ley (1932), whose independent experimentson the instrumental conditioning of headmovements of guinea pigs closely parallelssome of Skinner’s early operant conditioningwork, described in detail the use of a kymo-graph to record the head movements of hissubjects, albeit not as cumulative responses.Others in physiology and psychology, howev-er, recognized the utility of a cumulative ky-
mogram and developed early prototypes(Todd, 2004).
From Kymograph to Cumulative Recorder
Skinner first used a kymograph like theHarvard Instruments Company model shownin Figure 4 to record the behavior of a wholeorganism, an intact young rat, as it tuggedagainst a torsion wire to which its tail was at-tached (Figure 2 in Skinner, 1956). Soonthereafter, he began experimenting with un-restrained rats running through a longstraight tunnel, and later through a continu-ous rectangular tunnel, placed on a fulcrum(Figures 4 and 5 in Skinner, 1956). Each timethe rat passed over the fulcrum the tunneldropped, moving the stylus and recording aresponse on the smoked paper on the rotat-ing kymograph drum. From the figures, re-sponses were apparently recorded as simpledeflections of an otherwise straight horizon-tal line (cf. Figure 6 in Skinner, 1956). By1930, he was recording responses cumulative-ly, and published his first cumulative records,shown in Figure 5, in that year (Skinner,
334 KENNON A. LATTAL
Fig. 5. The first published cumulative (kymographic)records of Skinner (1930). The notations ‘‘N’’ and ‘‘t’’refer respectively to cumulative response number andtime.
Fig. 6. Diagram of reset mechanism devised by Skin-ner (1933) for the kymograph. See text for details.
1930). He later described the cumulative re-cording of contacts with a feeding device:
In recording the contacts a simple signal lever[stylus] writing upon a kymograph drumwould suffice, but we are interested in rate andin changes in rate, and a slight modificationin the method of recording simplifies the in-spection and measurement of these aspects ofthe record. A writing point is mounted upona small collar, which slides freely upon a ver-tical shaft. The collar is supported by a silkthread, which is wound about the shaft of aratchet mounted at the top of a vertical shaft.The ratchet is turned one notch at each con-tact by a coil magnet. Thus, when the circuitto the coil is made, the ratchet turns onenotch, the thread is wound slightly furtherabout the shaft of the ratchet, and the writingpoint is drawn up a corresponding (and uni-form) distance. The record of a series of con-tacts is a step like line traced diagonally uponthe kymograph paper. The slope of the linevaries with the frequency of the contacts, andthe line is straight, of course, if the rate is
constant. The record may be treated as agraph . . . (1932, pp. 26–27)
Although this modification of the kymo-graph solved the problem of cumulating re-sponses, a second problem remained: Whathappened if the number of responses in asession exceeded the relatively narrow verti-cal width of the paper? Some type of resetdevice was needed to return the pen to thelower edge of the paper. Skinner (1933) iden-tified and solved the problem as follows withthe device shown in Figure 6 (which is Figure3 from Skinner, 1933):
In the case of large amplifications of the ver-tical movement of the writing-point an im-practical width of kymograph paper isrequired. The effective width of the usual 6’’paper may be increased ten- or twenty-foldwith the device represented in Figure 3. Thespool (A), mounted on the sliding carriage ofthe writing-point (B) carries a supply of extrathread. The spool is held from unwinding bythe arm (C) and the catch (D). The entire de-vice is supported by the thread (F) which goesto the [response device]. As the carriage isdrawn up, a string (E), fastened at one end tothe lever (D), and at the other to the base ofthe stand, is drawn tight. The string is of sucha length that, when the writing-point reachesthe top of the drum, the lever (D) has been
335HISTORY OF THE CUMULATIVE RECORDER
Fig. 7. One of a bank of 24 ‘‘continuously recordingunits’’ developed by Heron and Skinner (1939).
lowered to a point at which the arm (C) is re-leased. The spool (A) immediately unwinds.The whole device drops, and the lever (D)catches the arm (C) after only one revolutionof the spool. A length of thread equal to thecircumference of A is thereby introduced intothe system. If the circumference is slightly lessthan the width of the kymograph paper thewriting point is almost instantaneously shiftedfrom the top to the bottom of the drum. Theoperation is repeated when the writing-pointagain reaches the top of the paper. (pp. 12–13)
The problems of reliable stepping and re-setting of the response pen were not easilysolved, and as will be seen, remained trouble-some for the next two decades. Nonetheless,by 1933 Skinner had worked out his idea ofcumulative recording, articulated its signifi-cance in his approach to the psychology oflearning, and adapted the kymograph to hispurposes. In the course of the latter he iden-tified the problems of cumulating responses;and through the reset device describedabove, made possible continuous recordingof such responses across session times that ex-ceeded a single vertical sweep across the ky-mograph paper by the stylus.
REFINEMENT: I. THE SKINNER-KELLER AXIS
University of Minnesota
In Skinner’s early experiments, the record-ing instrument remained what appears to theeye to be a modified kymograph; that is, itstill looked like one of Ludwig’s drums, butwith some unusual, indeed, idiosyncratic, fea-tures. In his renewal application for his Na-tional Research Council Fellowship (at Har-vard), Skinner (1979, p. 86) described a ‘‘raterecorder’’ that ‘‘included a gang of fourdrums driven by a Telechron motor—one ofRalph’s [Gerbrands] more beautiful pieces ofequipment, much more reliable than the oldkymograph drums’’ (p. 86). No further de-scription of this rate recorder could be foundin his other writings, but a seemingly relatedinstrument was described in detail by Heronand Skinner (1939). They used a bank of‘‘continuously recording units’’ to cumula-tively record responses and mechanically av-erage them across platoons of rats tested con-currently in different chambers, thusproducing an average cumulative record.
One of the component recorders, shown in
Figure 7, more closely approximates in ap-pearance the familiar cumulative recorder ofthe late 20th century. Instead of a verticaldrum, the drum now is horizontal and theseemingly frail string system for stepping andresetting the kymograph response pen is re-placed by a more substantial worm-gear drivefor stepping the pen. Also, rather than a sin-gle sheet of paper attached to a large rounddrum, a continuous roll of 6-in. paper un-winds across a platen turned by a constant-speed motor. Unlike the earlier system of cu-mulating responses in the kymographinstruments, here the response pen wasstepped by a clutch-driven system such thateach response energized the clutch, whichgripped a plate affixed to a shaft. As men-tioned earlier, the instrument was designed tooperate in concert with 23 other units to pro-duce ‘‘average response curves’’ from groupsof rats. Figure 8 shows a photograph of theentire bank of 24 recorders. In his autobiog-raphy, Skinner called this recording instru-ment a ‘‘summarizer’’ (1979, p. 223), part of‘‘a monster of an apparatus,’’ (1979, p. 222),a ‘‘mechanized version of the statistician’sLatin Square’’ (1979, p. 223), and noted that‘‘the industrial revolution had come to thestudy of animal behavior’’ (1979, p. 222). The
336 KENNON A. LATTAL
Fig. 8. The bank of 24 ‘‘continuously recordingunits’’ developed by Heron and Skinner (1939).
Heron-Skinner summarizer remained in thePsychology Department at the University ofMinnesota until 1965, at which point it, likeso many pieces of obsolete psychological re-search apparatus, was discarded to makeroom for a new laboratory (B. Overmeier,personal communication, July 24, 2004).
A variation of the summarizer was de-scribed by Skinner as follows:
On Project Pigeon [circa 1940] we had impro-vised a kind of cumulative recorder in whicha pen was drawn across a moving strip of pa-per by a taut string. The ratchet driving thestring was made from an old gear. I thoughtit was time to standardize a recorder for gen-eral use; graphs from different laboratorieswould then have the same dimensions. . . . Ifound a small standard circular saw blade thathad the right number of teeth for a ratchet,and I designed a paper drive mounted on achassis of sheet aluminum. Unfortunately thesaw eventually cut into the pawl that drove it,as I might have expected, and there were oth-er problems. I asked Fred [Keller] if hethought that we could get firm orders fortwenty or thirty instruments (from Yale, say, aswell as from Columbia and Indiana) to en-courage a company to develop a good model.I was soon reporting that a company was will-ing to go ahead without committed orders,and Fred and I eventually tested a prototype.It proved to be not only expensive but inac-curate. As in our behemoth, the pin (sic) wasdriven by a screw, and when we operated it ata steady rate, it drew a slightly wavy line. (1979,p. 322).
The above description seems to have been
of events when Skinner was at Indiana Uni-versity (1945–1948), an observation also sug-gested by the fact that some orders for thecommercial version were expected from In-diana. During this period there are two otherpieces of evidence in the history of the cu-mulative recorder that may or may not be di-rectly related to the instrument Skinner de-scribed above.
The first is a famous photograph of Skin-ner taken in his laboratory at Indiana Uni-versity in 1948, reproduced in Figure 9. Thefocal point of the picture is Skinner and apigeon in the foreground; however, in thebackground, behind Skinner’s right shoulder,are what looks to be a bank of three record-ing instruments. An enlarged portion of thephotograph’s background appears in Figure10. Several things about the recorders remainmysterious, however. They seem to be orga-nized into an interconnected bank, like thosedescribed previously by Skinner (1979, p. 86);however, there appear to be three, not fourin the bank as Skinner described. A drum ishard to see in the instruments, but the rightedge of the paper in the recorder in the fore-ground appears to be rounded at its right rearedge, perhaps suggesting a drum under thepaper. Furthermore, the instruments are notin an aluminum chassis, but are simplymounted on a table. Thus these instrumentsdo not appear to be the prototype referredto in Skinner’s preceding description. Thephotograph is not of sufficient resolution toassess the stepping or paper drive mecha-nisms.
The second piece of information is an ad-vertisement for a ‘‘cumulative recorder’’ thatappeared in the ‘‘Conference on the Exper-imental Analysis of Behavior Notes’’ datedJune 26, 1948, which stated that ‘‘several stan-dard pieces of apparatus can be supplied byMr. E. E. Lane, c/o Department of Psychol-ogy, Indiana University.’’ A list followed thatincluded the following description of a cu-mulative recorder:
Cumulative Recorder. Aluminum chassiswith endless chart movement and paper spool.Takes any width of paper up to 6 ½0. Paperspeeds of 40 (rat) or 80 (pigeon) per hour ap-proximately (others to order). From one tofour pen movements lying 80 apart can beused with one chart drive. Pen is driven by aratchet, designed with low mass and long
337HISTORY OF THE CUMULATIVE RECORDER
Fig. 9. Skinner in his laboratory at Indiana Universityin 1948.
Fig. 10. Enlargement of the photograph in Figure 9,expanding the portion over Skinner’s right shoulder toreveal a bank of three instruments.
stroke of armature to give highly reliable re-cords with variable lengths of contact. Inter-changeable scales of approximately 100 and200 responses per inch.
CUMULATIVE RECORDER—chassis of ap-propriate size, chart drive, and paper spool.(Specify speed)One Pen Unit $27.50 Three Pen Units $58.50Two Pen Units 43.00 Four Pen Units 75.00
The date and description make this record-er sound like the prototype Skinner and Kel-ler were working on; however, the above in-strument is advertised as giving ‘‘highlyreliable’’ records but Skinner had alreadycommented on the unreliability of the pro-totype. This unreliability, however, may havebeen manifest after the advertisement waswritten. The fact that the recorders are avail-able in units of one to four suggests a simi-larity with the recorders shown in the 1948photograph (Figure 8). The description ofthe chassis is not sufficient to relate it to theone in the photograph.
J. Dinsmoor (personal communication,June 20, 1998; note that all further citationsof Dinsmoor, unless preceded by an initial,refer to J. Dinsmoor) inspected the photo-graphs shown in Figures 9 and 10 and com-mented as follows:
When I came to Indiana 3 years after Skinnerleft , I commandeered an apparatushaving a single paper drive and four recordingmechanisms, lined up serially to produce fourindividual records during the same time peri-od. It had all the marks of BFS mechanicalingenuity. Each pen rested on a metal plate,
which was trapped by a notch cut in a fiberbelt. When the carriage reached the top of therecord, it rode up an incline that freed it fromthe current notch, and a weight on a stringpulled it back to the baseline, where it wascaught by the next of three notches on thebelt. In short, it reset! There was also someprovision, I believe, for displacing the pen bya couple of millimeters to leave a pip belowthe line or to distinguish portions of the re-cord by their level.
Concerning the pip, Dinsmoor (personalcommunication October 1, 2004) noted:‘‘Pips were made by displacement of the penwith respect to its carriage, but . . . these in-stances were actuated by hand rather than byelectric current.’’ Sample cumulative recordsproduced by this recorder can be viewed inthe Journal of the Experimental Analysis of Be-havior (JEAB) Volume 1, p. 146; Volume 5, pp.43 and 44; and Volume 9, pages 548 and 549.Although features of this recorder could besimilar to the one in the photograph in Fig-ure 10, on further reflection Dinsmoor lateradded (personal communication, June 23,1998):
I do not think the stuff in the picture couldbe the recorder I had. The one I used wasmounted on brass legs, had a motor under-neath, and unrolled a roll of paper. Unless itwas a great camouflage job, I don’t see howall of that could have been hidden in the pic-ture.
In summary, although details of the cu-mulative recorders developed at Indiana are
338 KENNON A. LATTAL
Fig. 11. Anger’s diagram of details of a cumulativerecorder constructed from parts given to him by Skinner(D. Anger, personal communication, May 20, 1999).
missing, from the available evidence it can beconcluded that Skinner experimented withtwo aspects of the recorder during this peri-od: A taut string was driven by a ratchet (sim-ilar to the ultimate solution eventually mani-fest in the Gerbrands recorders), replacingthe screw or worm gear type mechanism thatwas used in Heron and Skinner’s (1939) be-hemoth. Sometime in this period a fiber beltto guide the response pen as described aboveby Dinsmoor (personal communication, June20, 1998) also was used. The clutch-driven re-set for the response-pen unit described byHeron and Skinner also appears to have beenreplaced by the channel-release mechanismDinsmoor described above (personal com-munication, June 20, 1998), a mechanismthat will appear again in the Gerbrands Mod-el C-1 recorder.
Skinner left Indiana University and re-turned to Harvard University in 1948. Workon the cumulative recorder continued there.Douglas Anger (personal communication,May 20, 1999) described his experience withan early cumulative recorder mechanism giv-en to him at Harvard by Skinner (see Figure11 for Anger’s own diagram of the device):
When Skinner came [to Harvard] he gave mesome parts of an early recorder of his: a biggear, I guess about 60’ in diameter, with lots ofteeth, a solenoid, a big nail attached to a metalpiece, and a leaf spring as shown. It was de-lightfully crude but effective. I used that ratch-et mechanism to power a [cumulative record-er] I built. I used a round-and-round designto avoid the problem of resetting. A threadwith three pens attached went round and
round so that as one pen left the paper anoth-er one came on.
As noted above, Dinsmoor recalled that therecorders he inherited from Skinner at Indi-ana had a device with which one could man-ually make a pip to mark reinforcement. An-ger (personal communication, May 20, 1999)offered the following addition to the historyof the origin of the reinforcement markingor pip function of the response pen:
The recorder sat on a table with my circuitthat was recording IRTs [interresponse times].The IRT gadget was reset with a heavy weightthat shook the table, and caused a tiny move-ment of the pens on the thread at the time ofa response. I noticed those marks, and since Iwanted to know which IRTs were reinforced,I mounted a relay to hit the thread, and makea good mark when a reinforcement occurred.The first time I went in to Skinner with thoserecords, everything was so typical of Skinner.With just a glance at the records and before Icould say a word, he points to the pips andsays, ‘‘What’s that? What’s that?’’ I explained.Then nothing more was said about the pips.Now I really don’t know how fast it was, but itseemed like the next day I went into his lab,all his recorders had relays kicking the pensto make such marks.
After Skinner’s return to Harvard, he wassoon joined by Charles Ferster, who was com-pleting his third year in the doctoral programat Columbia. Morse (personal communica-tion, October 23, 1998) recalled the pivotalrole of Ferster in developing the instrumen-tation associated with operant conditioningin the 1950s and 1960s, including bringingthe ubiquitous snap lead and the Sodeco dig-ital event counter into the operant laboratory(see also, Ferster, 1953, 1970). The details ofthe development of the cumulative recorderfrom the time Skinner returned to Harvardand the emergence of the Gerbrands ModelC-1 are murky and may be forever lost; how-ever, at least a part of that history can be re-constructed. According to W. Morse (person-al communications, May 27, 1995; October23, 1998) the team of Ferster, Skinner, andRalph Gerbrands designed and constructedfour variations of the cumulative recorder,the last of which became the Model C-1, thefirst successful commercial version. A set ofinstructions for the Model C-3, written by theRalph Gerbrands Company, described the
339HISTORY OF THE CUMULATIVE RECORDER
previous models as A, B, C, C-1, and C-2.Thus the four models mentioned below inthis section will be labeled A, B, C, and finally,C-1.
According to Morse (personal communi-cations, May 27, 1995; October 23, 1998), thefirst model (A) was in the shape of the laterGerbrands commercial recorders and had aroll of paper that fit in it as it did in the C-1.The response pen was reset by the release oftension; however, Morse could not recallwhether the mechanism was a spring or grav-ity driven (in light of Dinsmoor’s commentof June 20, 1998, it seems reasonable to spec-ulate that the lead-weight/ gravity mechanismis the more likely candidate). This modelused a Ledex rotary switch to step the re-sponse pen, but the switch proved no matchfor the large number of responses that wereoccurring, causing Ferster to recall that ‘‘fora long while, I spent much of my time re-placing and repairing rotary solenoids whichlasted for only a few hundred thousand op-erations’’ (1970, pp. 39–40). According toMorse (personal communication, October23, 1998), the second model (labeled the‘‘Brown Box’’ Model by Morse) was housedin a large brown box because it had a verylarge reciprocating motor. It had an Auto-matic Electric stepping switch to step the re-sponse pen (which ultimately found its wayinto the Gerbrands Model C-1 and subse-quent models) and two leather-like frictionplates pushed together as the pen stepped upthe page. These plates eventually began toslip, rendering this solution to stepping thepen unreliable; the mechanism sounds relat-ed to the one described by Heron and Skin-ner (1939).
The reset mechanism, which was con-trolled by ‘‘a powerful [reciprocating] motorthat overcame the friction on these plates,’’was such that ‘‘it spun the disc back to thebottom of the page’’ (Morse, personal com-munication, October 23, 1998). One advan-tage of the friction plate device, according toMorse, was that the response pen could bereset at any point on the cumulative record—the previous versions did not allow for this.This function was especially useful for analyz-ing certain schedules, such as the fixed inter-val (Gollub, personal communication, August3, 2004).
Skinner (1979) later summarized these ear-
ly attempts to develop a reliable cumulativerecorder at Harvard as follows:
The taut-string recorders were not workingwell. The ratchets jammed, the paper creptsideways, pens skipped when moving fast andleft blobs of ink when standing still, and welost many valuable records. Charlie [Ferster]and I designed a new model but ran into an-other problem. The stepping switch whichmoved the pen was rated for only a few hun-dred thousand operations and pigeons re-sponding at high rates for as much as tenhours a day soon wore it out. We eventuallyfound a switch that would operate many mil-lions of times and, meanwhile, . . . Ralph Ger-brands was working on a better model with aTelechron motor and a pen that reset when itreached the top of the paper. (p. 38)
The taut string recorders either were identi-cal, or closely related, to the first model de-scribed by Morse. Presumably, the third Ger-brands version (the model labeled C in theGerbrands instruction sheet mentionedabove) eventually evolved into the Model C-1, though the details of the C model are notknown. ( J. M. Harrison [Personal communi-cation, October 23, 1998] gave a very early‘‘prototype’’ Gerbrands recorder to the Har-vard University instrument collection thatmay have been a Model C.) An advertisementin Volume 1, Number 3 (August, 1958) ofJEAB is for a Gerbrands ‘‘Model C cumulativerecorder.’’ This cannot, however, refer towhat is here described as the prototype Mod-el C for three reasons. First, the Model C-2had replaced the Model C-1 by this time. Sec-ond, an advertisement for the ‘‘patentedModel C’’ appears in July 1960, but based onthe patent information this refers to the Mod-el C-2, though it is not described as such inthe advertisement. Third, in January 1961,the Model C-3 was introduced in an adver-tisement by noting ‘‘This is to announce ournew model C-3 cumulative recorder–succes-sor to the model C-2.’’ Three model chang-es—C to C-1, C-1 to C-2, and C-2 to C-3—could not have occurred between July 1960and January 1961. The C-1 and its descen-dants will be described in detail shortly (fora synopsis of Gerbrands’s contributions to theinstrumentation of behavior analysis, see K.Dinsmoor, 1987).
Two other pieces of anecdotal evidencefrom this period may be added to the above,
340 KENNON A. LATTAL
although the details are too sketchy to allowany firm assertions about how these pieces fitinto the history of the Models A through C-1. As a graduate student in the mid 1960s atthe University of Alabama, I recall in the op-erant laboratory there was an early-model cu-mulative recorder that reportedly wasbrought to the University by Gilbert (see, e.g.,1958) when he came to teach at Alabamafrom Harvard in the 1950s. The recorder wasabout the same size as the later ones devel-oped by Ralph Gerbrands and was housed ina wooden box. It had a platen made of brass,the response pen was moved up the paperalong a string, and ‘‘the other end of thestring hung overboard over a little pulley andhad a weight attached to provide the resetaction’’ (W. Sullins, personal communication,May 18, 2003). Sullins added that ‘‘I remem-ber having the sense, from the quality of itsconstruction and operation, that it was not a‘‘one of a kind’ production.’’ The existenceof more than one such recorder was corrob-orated by M. Harrison (personal communi-cation, November 11, 1998), who also re-called using a recorder ‘‘made of wood—varnished—with brass platen.’’ Theserecorders could be variations on the Model Adescribed above, but it is impossible to con-firm this given the absence of the recorderitself, which disappeared from Alabamasometime in the 1980s. (Harrison also notedthat the version of this recorder that he usedstepped from right to left rather than thestandard left to right arrangement. Examplesof the latter records may be seen in Harrisonand Abelson [1959, p. 33, Figures 11, 12, 14,and 16].)
The second item is a recollection of L. Gol-lub (personal communication, July 26, 2004)concerning a cumulative recorder that he sawduring his graduate study at Harvard in thelate 1950s: ‘‘[It] was built inside a lovelywooden box with a glass top. Reset was ac-complished by a reciprocating arm thatpulled the pen carriage back to the startingposition. This model did not have a clutch.’’He later added (personal communication,August 2, 2004) that ‘‘events could bemarked on the cumulative response recordwith a short horizontal line produced by asolenoid.’’ This recorder sounds like theModel B described by Morse above, but likethe wooden box recorder described by my-
self, Sullins, and Harrison, it is not possibleto place this recorder unequivocally in theGerbrands schema of Models A through C-1.
As with the Indiana years, even though elu-cidating all the details of the developmentalhistory of the cumulative recorder at Harvardbetween 1948 and the early 1950s cannot bedone, the available evidence suggests a peri-od of very active and continual experimen-tation with both the response-pen unit stepand reset functions. This experimentationculminated with the successful constructionof a durable and reliable model that met thechallenges of intensive use during the hugeresearch program on schedules of reinforce-ment of Ferster and Skinner (1957). Beforedescribing these final steps, some other de-sign efforts along the way must be men-tioned. These were conducted at a differentvenue by Fred Keller and his colleagues.
Refinements Influenced by the ColumbiaGroup
Concurrent with Skinner’s refinements ofthe cumulative recorder, his friend and col-league from graduate school days, Fred Kel-ler, also was working on versions of cumu-lative recorders. In a colloquium atNortheastern University, Keller (1975) re-called his departure from Colgate Universityto teach at Columbia University, noting thathe departed
. . . in the post-depression period, after sevenlean years at Colgate University, without pro-motion or a raise in pay, and I brought myresearch equipment with me: A Skinner boxand a cumulative recorder. Footnote: TheSkinner box was early Gerbrands, described inthe B of O [Skinner’s Behavior of Organisms],and the cumulative recorder, made by a friendin the Physics Department, a brother-in-law ofthe psychologist Douglas Ellson, was builtfrom a Harvard kymograph drum, a Tele-chron clock motor, two Boston gears, a singleball bearing, four supporting parts, a 2-footbrass rod, a piece of silk thread, and a Bristolpen. My yearly research fund at Colgate was25 dollars—which encouraged economies.
Frick, Schoenfeld, and Keller (1948) sub-sequently published a description of the ‘‘Ap-paratus designed for introductory psychologyat Columbia College’’ that included a pho-tograph and description of the adapted ky-mograph. A published photograph of James
341HISTORY OF THE CUMULATIVE RECORDER
Fig. 12. Photograph of (left to right) Rodolfo Azzi,Fred Keller, Maria-Amelia Matos, Carolina MartuscelliBori, and Andres Aguirre with a kymograph-based cu-mulative recorder, December, 1961.
Fig. 13. Prototype cumulative recorder developed byFred Keller in conjunction with Bill Bell.
Fig. 14. View of the cumulative recorder shown inFigure 13 with the cover removed.
Dinsmoor as a graduate student at Columbia(Figure 3 in Dinsmoor, 1990) shows such akymograph (directly behind the operant con-ditioning chamber in the photograph) in use.The instrument was elegantly simple: an elec-tric motor turned the drum, which wasplaced on its side. Responses were marked onthe paper attached to the drum by a pencilin a holder. The pencil was pushed up thepage by connecting its holder to a worm gearthat rotated with each response. When Kellerwent to Brazil to help organize programs inpsychology, a larger version of the kymo-graph-type cumulative recorder, with a verti-cally mounted drum and automatic reset de-vice appeared, but according to those whoused it, it was not very reliable (M. H. Hun-ziker, personal communication, October 27,1998; see Figure 12).
Keller also was responsible for a prototyperecorder labeled the Teaneck recorder by D.Eckerman (personal communication, Sep-tember 14, 1998), which Eckerman describedas ‘‘something that [Fred Keller] inventedand talked some local talent into making aprototype of.’’ This prototype model is shownin Figures 13 and 14. According to Keller’sson, John Keller (personal communication,November 11, 1998), the Kellers lived nextdoor to an engineer named Bill Bell around1949–1950. Fred Keller described the cumu-lative recorder to him and discussed the needfor an inexpensive instrument (perhaps fol-
lowing the failure of the expensive version de-scribed by Skinner, 1979, p. 322) with Bell,who, in turn, was looking for new opportu-nities. Bell reportedly designed the recorderand built several copies of it, but John Kellerdid not know of their fate. Dinsmoor (per-sonal communication, October 1, 2004),however, recalled that ‘‘six copies of the Bellrecorder were purchased for use in an un-dergraduate laboratory at Indiana [Universi-ty]; I believe they used ball point pens, whichwere convenient to maintain but did not con-sistently leave a tracing on the paper.’’ Know-ing the identity of the person who construct-ed it and the dates, this instrument does notappear to be the same Skinner had describedas a joint effort with Keller to develop a com-mercially viable cumulative recorder (seequotation on p. 336 above).
342 KENNON A. LATTAL
Fig. 15. Three views of a Ralph Gerbrands Company Model C-1 cumulative recorder. Top: frontal view with thelid raised. Lower left: side view showing lead weight-reset mechanism at the top of the metal cylinder. Lower right:Detail of channel for disconnecting the response pen unit from the ladder chain. The long metal bar defines thechannel and the metal sprocket pushing the pen holder is just at the base of the channel.
REFINEMENT II. THE GERBRANDSCUMULATIVE RECORDERS
The Gerbrands Model C-1
The Ralph Gerbrands Company Model C-1 recorder was commercially available in theearly 1950s. The first sales figures availablefor it are from 1955 (K. Dinsmoor, 1987),
when 36 were reported sold. An advertise-ment for what is called a ‘‘Model C,’’ but wasprobably a Model C-1, appeared in the firstissue (1958) of JEAB, a year in which 242 cu-mulative recorders were sold (K. Dinsmoor,1987). Figure 15 shows some of the details ofthe Model C-1, many of which reflect refine-ments of earlier instruments developed by
343HISTORY OF THE CUMULATIVE RECORDER
Skinner as described above. The C-1 washoused in an aluminum chassis that heldboth the mechanical components of the re-corder and a roll of 6.5-in. wide paper thatwas fed continuously across a platen. A largeexternally mounted 110 volt AC motorturned the platen. Each operation of a step-ping switch pulled a pen holder one unit(ratchet step) up the paper for each re-sponse; the pen holder rested against a pieceof metal protruding upward from a ladderchain. As the pen holder moved up the page,it also pulled up a weight attached to theholder by a string. The lead weight was en-closed in a metal cylinder (see Figure 14, low-er left photograph). The reset mechanismwas a simple combination of mechanical andgravity arrangements so that as the pen hold-er approached the top of the paper, it movedthrough a channel that separated it from thechain. When this occurred, the lead weightwas free to descend and, returning to its rest-ing state, pulled the pen holder back to thelower edge of the paper. When it reached thelower edge of the paper, the pen holder im-mediately engaged another metal sprocketon the chain and as responding continuedthe chain was again pulled up the page andwith it the pen holder. This reset apparatusappears similar to that described earlier byDinsmoor (personal communication, June20, 1998) and by Morse (personal commu-nication, May 27, 1995; October 23, 1998).
Although the stepping switch for the penproved to be reliable through millions of op-erations, the step of the pen was not alwayseven and the pen-holder resetting device wasless than perfect. Unlike the Model B resetmechanism described above, the responsepen could not be reset except when itreached the top of the page. More impor-tantly, however, the pen mechanism some-times stuck and would not reset completelyto the bottom of the page before reengaginga metal sprocket. A solution to this problem,however, had to await the next model.
Gerbrands Model C-2
The Model C-2 appeared between 1952and 1956. A Foringer equipment cataloguedated August 25, 1956 included the followingrecommendation with a description of its cu-mulative recorder control panels: ‘‘We rec-ommend that you buy the Gerbrands’ Model
C-2 24 V. D.C. Cumulative Recorder.’’ The C-2 received U.S. Patent number 2,926,984.Perhaps the most important change in ap-pearance from the C-1 to the C-2 was the re-placement of the thick ladder chain that en-gaged the response pen holder first with astring in a prototype model (Figure 15, bot-tom left photograph) then, in the final ver-sion, with a smaller bead chain to which theresponse pen holder was attached (Figure 16,top right photograph). The change to thesmaller bead chain was part of a major mod-ification in the method of resetting the re-sponse pen. In the C-2, the earlier C-1’s grav-ity-based reset mechanism was replaced by aclutch-based reset system. When the responsepen holder reached the top of the paper amicroswitch activated a solenoid. The sole-noid in turn engaged a clutch that allowedthe response pen holder to be returned tothe bottom of the page by a spring-drivenmechanism. The latter was inspired by the de-vice that reset a rotary telephone dial (Morse,personal communication, October 23, 1998).Gerbrands obtained a patent on this clutch-reset mechanism in March of 1960, giving hiscompany ‘‘a competitive advantage over theother companies selling this type of recorderat that time’’ (R. Gerbrands, letter to W.Street, 1990; see also Street, 1960). Like theC-1, the C-2 had a wide chassis (31 cm, notincluding the motor) and a large Telechronmotor on the outside of the chassis to drivethe platen. Here is a detailed descriptionfrom the instruction manual accompanyingan early C-3 recorder:
The cumulator [response] pen is driven acrossthe paper by a high speed, electromechanicalstepping device. The reliability and durabilityof this device has been demonstrated by itslong use in automatic telephone exchange ser-vice. It will respond to a maximum rate of 35to 50 pulses per second. Its life expectancy isapproximately 50 million pulses [responses].As used in our recorder it has exceeded thisfigure by many fold. The angular movementof the stepper is transmitted to the pen systemthrough a gear train and a non-slip electro-mechanical clutch. A chain wrapped aroundthe drum of the pen drive system, advancesthe pen along the paper. The reset force isprovided by a governor controlled motor. Thismotor is spring powered and returns the pen,without bounce, to its zero position at a rapidand constant speed. The clutch is one of the
344 KENNON A. LATTAL
Fig. 16. Four views of early Gerbrands Model C-2 cumulative recorders. Top left: frontal view with the cover raised.Top right: top view showing chain drive. Bottom left: top view of a Model C-2 prototype that used a string ratherthan a chain to pull the response pen unit (Note: the event pen visible on extreme right of the figure was added byan investigator). Bottom right: Detail of reset clutch mechanism assembly. The solenoid below the bar disengagesthe step mechanism above the bar, and allows the pen to reset. The Automatic Electric stepping switch coil is visibleat the lower left.
essential features of our recorder. In compar-ison with the clutch used in an electric chron-oscope which reads to only .01 seconds, itneeds to be and is more than 10 times as ac-curate. Differential and tooth-wheeled clutch-es are not sufficiently accurate for use in cu-mulative recorders.
At some point, reportedly due to a laborstrike that halted availability of the largeTechtron motors, Gerbrands began using asmaller Cramer motor, mounted on the in-side left side of the chassis. This motor, or avariation of it, was used on this and all sub-sequent models of the recorder. The smaller
motor also made it easier to change the gearsdriving the platen, and thus the speed of theplaten (changing gears was an option on allof the Gerbrands commercial recorders). Aphotograph of this latter version of the modelC-2 with the smaller motor located on the in-side of the chassis is shown in Figure 17. Fi-nally, the C-2 also had a mechanism, locatedon the drum of the pen drive system, thatallowed the user to set the height that thepen would reach on the paper (i.e., the num-ber of responses) before it would automati-cally reset to the bottom of the paper (see leftwheel in the top right view of the recorder in
345HISTORY OF THE CUMULATIVE RECORDER
Fig. 17. A late Gerbrands Model C-2 (left), with the smaller Cramer motor located inside the chassis, and a ModelC-3 cumulative recorder (right). The motor is visible on the left inside of the C-2 and on the right inside ofthe C-3.
Figure 16). In addition, the pen could be re-set electrically by a remote pulse from theprogramming equipment. Interestingly, theC-2 was never advertised as such in JEAB, onlyas the ‘‘Model C.’’
Gerbrands Model C-3, C-3HS, and C-4
The most successful of all of the cumulativerecorder models of the Gerbrands Company,the Model C-3 was a remarkable machine(see Figure 17). Taking advantage of modifi-cations in the electromechanical configura-tion of the step/reset mechanism, in com-parison to its forerunners the C-3 had anarrower chassis (27.5 cm), gray in color(like the C-1 and C-2) in earlier units and tanin later ones (see Figure 18, top photograph,although the photograph is not in color).The height (22.5 cm, minus the rubber feet)and depth (17.2 cm) were the same as the C-1 and C-2. The stepping and reset mecha-nisms were similar to the Model C-2, but dif-fered in some important ways as noted below.The C-3 was first advertised in JEAB in Janu-ary 1961 and for the last time in July 1987, adistinguished 26-year run. So durable is itthat many of these C-3s are still in regular usetoday.
A set of instructions for the C-3 from theRalph Gerbrands Company distinguished the
C-2 and C-3 in terms of the ease of changing‘‘response rates,’’ referring to the size of thestep of the response pen. The gears respon-sible for this function on the C-3 were moreconveniently located than on the C-2 (see Fig-ure 18, lower left photograph). A button onthe top of the chassis also allowed manual re-set of the response pen, in addition to theautomatic and adjusting reset features of theC-2 (that is, this recorder, unlike the C-1,could be reset at any point in the responsepen’s traverse of the paper). In 1968, a ModelC-3 was priced at $375.00 and the Model C-3H.S. (described in the next paragraph) waspriced at $515.00.
In April 1963, an advertisement for a Mod-el C-3 H.S. (Sometimes referred to as ‘‘C-3S.H.S.’’ [‘‘silent high speed’’]) also appearedin JEAB. This was essentially the same record-er as the C-3 except that it used solid-stateelectronic circuitry, which made it operatefaster (more precisely tracking very high-rateresponding) as well as silently. Many who hadbecome accustomed to the reassuring click-ing sound of a C-3 as responses were beingemitted in the operant chambers found thesilence an unsettling change. Investigatorslearned to discriminate the sound of theirown experiments from the din of other op-erating recorders in the laboratory, and the
346 KENNON A. LATTAL
Fig. 18. Gerbrands Model C-3 recorders (with covers raised). Top: An early Model C-3 (left) and a late Model C-3 (right). Note the difference in chassis shape. Bottom left: top view of an early Model C-3. Bottom right: Detail ofan early Model C-3 showing the gears for stepping responses located behind the platen (foreground) and above thewhite roll of cumulative recorder paper. See text for details of this change from the Model C-2.
loss of this monitoring function was some-times significant. Wolkomir (1996) made asimilar observation with respect to commonhousehold appliances:
Consider the clunking sound made by someclothes washers as they open and close valvesto let water in and out. Those clunks can beannoying, and some manufacturers have, infact, eliminated them. But it turns out that theclunks tell users when the machine is chang-ing cycles. ‘‘My wife’s new washing machinedoesn’t clunk, so they had to add a buzzer totell you when the cycle changes,’’ explained[an expert on noise abatement]. (p. 56)
Indeed, researchers using computers to gen-erate cumulative records have been known to
attach auditory generators to response out-puts to allow this type of monitoring.
The last Gerbrands model, the C-4, was in-troduced in a JEAB advertisement in January,1989 (see Figure 1). A dial was added to thisversion on the back of the recorder that al-lowed specification of response pen steps permm of 2, 4, or 8. The recorder also operatedon solid-state circuitry and therefore did notmake noise when stepped. Otherwise it wassimilar to the later Model C-3s in appearance(tan chassis), but with the C-4 the AC motorswitch was on the side of the recorder ratherthan in the rear. This left room for the switchto adjust the pen step size to be attached tothe rear.
347HISTORY OF THE CUMULATIVE RECORDER
PROLIFERATION AND MODIFICATION
Cumulative recorders proliferated in twoways. Once it became apparent that a marketexisted for cumulative recorders, other com-panies created their own versions. In addi-tion, individual teachers and scientists modi-fied commercial cumulative recorders toaccommodate their special requirements.
The Gerbrands cumulative recorder re-mained the standard by which all the otherswere judged. In fact, several of the prominentelectronic equipment companies (e.g., Le-high Valley Electronics, the Grason-StadlerCompany, and Foringer) simply sold the Ger-brands cumulative recorders rather than con-struct their own versions; however, LehighValley Electronics did market a student mod-el, described below.
Figures 19 through 27 show the cumulativerecorders of other manufacturers of these in-struments from the 1960s through the 1990s.The recorders are shown in the order of thedates (indicated) of their first advertisementin JEAB. These dates seem a reasonable ap-proximation to the inception of these instru-ments because JEAB’s readers were the mostlikely people in the psychology community topurchase them. Two companies, however, didnot advertise their cumulative recorders inJEAB, Davis Scientific Instruments Companyand the Lafayette Instrument Company, al-though both companies did advertise otherproducts there. The prices of the recordersvaried considerably and a price for a givenyear (where catalogues providing prices havebeen found) is indicated to give some contextfor evaluating the relative costs of the differ-ent instruments.
The first recorder of another manufacturerto appear in a JEAB advertisement after theGerbrands Model C-1 was in January, 1962,when the Scientific Prototype Company ad-vertised its Model CR2D cumulative recorder,shown in Figure 19. The chassis of this instru-ment mounted conveniently on a relay rackand, unlike the Gerbrands C-1, did not re-quire a separate programming panel, becausethe studs to which snap leads attached to con-trol the functions of the recorder to operatewere on the face of the chassis itself. Detailsof the stepping mechanism are shown in thelower left photograph and described in thefigure caption. The cost listed in a July, 1965
advertisement in JEAB was $295.00. In July1962, the BRS Electronics Company adver-tised a modified Varian recorder as a cumu-lative recorder, as shown in Figure 20. Fewdetails are available about this recorder. Itsprice, noted in the advertisement, was$1,095.00, very expensive for the time.
In addition to the companies noted abovethat sold the Gerbrands recorders in theircatalogues, at least one of the other recorderswas marketed by different firms, sometimesunder the firm’s name and sometimes underthe name of another firm. The Campden In-struments Company of England, the StoeltingCompany of Chicago, and Behavior SystemsInternational of New York City all sold ver-sions of the same solid-state circuitry cumu-lative recorder. The Campden Instrumentsrecorder is shown in Figure 21. A version ofthis recorder was first advertised in JEAB inMarch of 1972 (by Stoelting). The StoeltingCompany advertised this same recorder, withthe label ‘‘Campden Instruments’’ on itsfront, and it was included in the company’s1974 catalogue. This company was still sellingthe same recorder in 1984, but by then with‘‘Cumulative Recorder 624’’ replacing the‘‘Campden Instruments’’ label. The cost of aBehavior Systems International recorder inan advertisement in JEAB in September, 1973was $489.00 for a recorder in a freestandingcase (shown in the photograph) and $459.00for one that could be mounted in a relayrack.
Figure 22 shows the Davis Scientific Instru-ments entry into the cumulative recordermarket. The Davis recorder was unusual inseveral ways. It had an aluminum and heavyplastic frame and chassis and included a dig-ital counter (see the bottom photograph). Inaddition, unlike most of the other recorders,it used a roll of 5-in. paper that was perforat-ed on both sides, similar to that used in Es-terline-Angus event recorders of the time. Acatalogue from the 1960s (exact date un-known) listed the recorder price as $412.50.Figure 23 shows two Lafayette InstrumentCompany cumulative recorders, one with ablue and the other with a gray chassis, butotherwise identical. The blue one was circa1985; the date of the other is unknown. TheLafayette recorder was advertised in its 1975catalogue, but not in JEAB. The cost in 1975was $415.00. Conversations with numerous
348 KENNON A. LATTAL
Fig. 19. Three views of a Scientific Prototype Company Model CR 2D cumulative recorder. Top: frontal view withcover down. Lower left: top view showing the worm-gear mechanism for stepping the response pen unit. The frontof the recorder is at the top of the image. The rod nearest the bottom of the image holds the roll of paper andabove it on the right is the ink reservoir. The worm gear is the rod just above the ink reservoir. Lower right: viewwith front cover raised, showing the stud panel for connecting snap leads to enable the different functions. Thebottom edge of the front cover (not shown) has a serrated edge for ease of cutting off the paper (the serrated edgecan be seen in the top photograph of this figure).
users of all of these non-Gerbrands instru-ments over several years have yielded a con-sensus that none of these instruments mea-sured up to the Gerbrands recorders fordurability and reliability.
Figure 24 shows a cumulative recordermanufactured in Japan. Interestingly, Skinnernotes in his autobiography that as early as1952, Gerbrands sent ‘‘a lever, a food maga-zine, and a kymograph for obtaining cumu-lative response records to a Japanese psy-chologist, paid for by the Japanese
government’’ (1979, p. 38). The cumulativerecorder is similar in shape and size to theGerbrands models, but is much heavier. Theelectronic components have Japanese char-acters written on them, but the functions ofthe recorder are written in English. There areno marks to identify the manufacturer.
Once a technology is available, novel ex-trapolations of its use beyond its exact origi-nal purpose often develop (cf. Diamond,1999), and so it was with the cumulative re-corder. One logical extension of its role in
349HISTORY OF THE CUMULATIVE RECORDER
Fig. 20. The first advertisement in JEAB for the BRSElectronics cumulative recorder ( July, 1962).
Fig. 21. A Campden Instruments/Stoelting Compa-ny/Behavior Systems International cumulative recorder,first advertised in JEAB by Stoelting Company in March,1972. The example shown was designed for use in GreatBritain (note the ac plug to the right of the recorder).
obtaining data from experimental research isto use it as a didactic tool in the classroomlaboratory. As noted previously, Keller and hiscolleagues at Columbia University developeda simple kymograph-based cumulative re-corder for student use. A recorder almostidentical in appearance to this one was latermade available commercially by the LehighValley Electronics Company (Figure 25) andwas advertised in the July, 1967 issue of JEABfor $145.00. From 1966, Gerbrands also pro-duced a relatively simple student recorder(Figure 26) consisting of a timing motor thatturned a worm gear, propelling itself alongthe x-axis of the paper. A response pen waspropelled along the y-axis by having respons-es rotate a mechanism inside the holder thatwas connected to a second worm gear thatrotated with each response, thereby drivingthe response pen holder up the paper. Thisrecorder sold for $195.00 in 1968. The DavisScientific Instrument Company produced astudent recorder that was somewhat similar inappearance to the Gerbrands student record-er (Figure 27). According to a catalogue de-scription of the instrument:
The recording pen is carried across the chartby a unique electromagnetic linear steppingmotor which moves itself along a pair ofsmooth parallel rods. The motor moves in dis-crete steps whose length can be continuouslyvaried in the range from 20 per inch to 300per inch by a simple dial adjustment . . . Theentire stepping motor and rod assembly is car-ried along the length of the record chart by asynchronous motor-driven screw to provide atime base.
The price of this recorder in the 1960s cata-logue mentioned above was $231.00. For aneven less expensive way of allowing studentsto generate cumulative records, van Sommers(1967) described how a manual typewritercould be converted into a cumulative record-er for $25.00.
In addition to creating new and relativelysimple instruments for didactic purposes,others adapted commercially available re-corders for idiosyncratic and specialized re-search purposes. For example, Hurwitz(1961), Kulli and Bogrow (1971), and Robles,Alucema, and Cohen (1981) each adaptedstandard cumulative recorders for recordingmultiple responses from different sources asmight occur under concurrent schedules ofreinforcement. In addition, J. Keller (1974)modified a cumulative recorder so as to pro-duce histograms.
A Note on Paper, Pens, and InkThe cumulative recorder was no better
than the records it produced, thus paper,
350 KENNON A. LATTAL
Fig. 22. A Davis Instrument Company cumulative recorder. Top: frontal view. The original response pen holderhas been removed and replaced with one suitable for a Leeds and Northrop glass reservoir pen (left). The eventpen holder (right) is original. Bottom: top view showing a digital counter that is part of the unit.
pen, and ink were important to producinghigh-quality records. Most of the recordersused similar rolls of 6.5-in. plain white paper,available from the instrument companies orother sources (see, e.g., Ator, 1991). As notedpreviously, one exception was the Davis Sci-entific Instruments Company recorders.
As described by Ferster (1970), the Leedsand Northrop glass reservoir pen became
standard even before the construction of theGerbrands Model C-1. It was used on all sub-sequent Gerbrands models. For a time, dis-posable felt-tipped pens were sold by the Ger-brands Company for their recorders. Thegreatest benefit of these pens, however, wasapparently to the seller and not to the sci-entist. These pens were short-lived, did notproduce as crisp lines as the Leeds and Nor-
351HISTORY OF THE CUMULATIVE RECORDER
Fig. 23. Two Lafayette Instrument Company cumulative recorders, same model numbers but different chassiscolors (left, blue; right, gray), presumably manufactured in different years. The paper runs under a serrated edgeto allow it to be cut off easily.
throp pens, and dried up quickly when leftuncapped. Other manufacturers, like Scien-tific Prototype, used a metal pen attached toan ink reservoir especially designed for theirrecorders. The Davis Scientific InstrumentsCompany cumulative recorder used a penwith a writing tip from a popular lettering kit(the K and E Leroy lettering kit). Snapper,Verhave, and Herman (1964) described an-other type of commercially available pen thatcould be adapted for use with the cumulativerecorder. Some investigators replaced thestandard pens on the Gerbrands recorderwith a polygraph-type pen with a larger inkreservoir. Finding just the right ink was thesubject of three early technical notes in JEAB(Lindsley, 1958; Russell, 1961; Verhave,1958).
Managing Cumulative Records
In the typical laboratory the cumulative re-corder generated kilometers of cumulativerecords that first had to be managed beforethey were removed from the recorder, lestthey clutter up the entire laboratory; andonce separated from the recorder, they hadto be conveniently and safely stored. To theseends, several instruments were constructed oravailable commercially. Various types of take-
up reels were available, including a LehighValley Electronics Model 481-03 for storingcumulative records as they rolled off the re-corder (see also Millner, 1970). Once re-moved, the records could be neatly folded us-ing a device created by Lekrone, Marmasse,Ferster, and Holtzman that folded records‘‘in the manner of an accordion pleat’’ (1963,p. 269). Emiley, Hutchinson, Hallin, and Kir-aley (1971) described a large cabinet de-signed specifically for the storing of records.
Selecting and Displaying ‘‘Representative’’Cumulative Records
Cumulative records were the primarymeans of data presentation in Skinner’s Be-havior of Organisms (1938) and for many yearsthereafter in research articles and texts de-scribing experimental analyses of behavior(see, especially Ferster & Skinner, 1957). Re-cords as data were presented in differentways. Sometimes, records from entire sessionswere shown, as illustrated in the top recordsin Figure 2. At other times they were tele-scoped, as noted in the introduction, by cut-ting out the space between the reset lines andcollapsing the records closer together to al-low more records to be presented in a givenamount of space. Telescoping of records is
352 KENNON A. LATTAL
Fig. 24. A Japanese manufactured cumulative record-er. Top: front view. Bottom: top view.
Fig. 25. Advertisement for a Lehigh Valley Electronicsstudent recorder, shown in the lower left corner. Thisrecorder is similar to the instrument described by Fricket al. (1948).
Fig. 26. A Ralph Gerbrands Company student cu-mulative recorder. The pen holder is angled to hold adisposable pen.
shown in the bottom records in Figure 2. Atstill other times, only small segments of a ses-sion would be presented, and sometimesmagnified, to illustrate a particular effect. Re-gardless of how the records were presented,it was difficult to choose from among thesometimes literally miles of cumulative re-cords generated during an experiment whichones were ‘‘illustrative’’ or ‘‘representative’’of the general performance during the ex-perimental conditions under discussion. Thecriteria for representativeness were left to theexperimenters’ own discretion and judgment.
In a poignant observation, the 1960s trioof Peter, Paul, and Mary sang of how for littleboys, ‘‘painted wings and giants’ rings makeway for other toys.’’ And so it was with therelation between researchers and cumulativerecorders beginning sometime in the heyday
of Peter, Paul, and Mary. Skinner (1976b, p.218) noted the passing of an era when he saidthat ‘‘Evidently we have not long to wait foran issue of JEAB without a single cumulativerecord!’’ Poling (1979) confirmed Skinner’sconcerns by demonstrating a precipitousdrop in the number of cumulative recordsper page in JEAB over the 20 years from 1958to 1978. Both the discipline and its technol-ogy were changing. Many years before Skin-ner’s lament, this change was enabled with
353HISTORY OF THE CUMULATIVE RECORDER
Fig. 27. Catalogue photograph of a Davis ScientificInstrument Company student cumulative recorder.
the widespread use of the digital counter asa means of accurately accumulating responsesover measured time periods. The laboriousand less reliable method of estimating re-sponse rates from slopes of cumulative re-cords therefore waned (see Carman, and Herrick  for examples of practicalsolutions for rate calculations from cumula-tive records). Precise quantification of re-sponding became the coin of the realm. Thetypes of quantitative analyses ushered in bythe seminal work of Herrnstein (e.g., 1970)and his students placed increasing emphasison aggregates of responses in time ratherthan on the more moment-to-moment anal-yses invited by the cumulative recorder.
Two caveats to this last observation shouldbe noted. First, digital counters in combina-tion with precise timing devices also greatlyfacilitated the interresponse time analysesthat epitomize much of what is sometimescalled molecular analysis. Second, much ofthe original rationale for using cumulative re-corders was in terms of allowing the analysisof rates of response. With respect to such ratecalculation, response accumulation on coun-ters was more accurate. Indeed, Killeen(1985) observed that ‘‘the sensitivity of therecorder [as a reflector of response rates] de-creases as the square of response rate’’ (p.179).
The digital computer began replacing elec-tromechanical equipment from the mid1960s forward (the first advertisement for aPDP8 computer-based experimental controlsystem appeared in the September 1972 issueof JEAB), thus it was simply a matter of timeuntil computer programs for generating cu-mulative records became widespread. Such
programs were not only more flexible andeasier to use, but cheaper than a cumulativerecorder. Some of the first commercial prod-ucts converted dot matrix printers to allowcumulative recording, and later softwarepackages became available to provide on-screen displays and storage of computer-gen-erated cumulative records.
The last Gerbrands company ad for a cu-mulative recorder appeared in the March1994 issue of JEAB. On September 30, 1994the Ralph Gerbrands Company closed itsdoors forever.
LIFE AFTER EXTINCTION
From its evolution from the kymograph toa unique scientific instrument in its ownright, the cumulative recorder and its prod-uct, the cumulative record, helped to shapethe experimental analysis of behavior beyondthe need for the instrument itself as a meansof data analysis. For more than 40 years, how-ever, from the 1950s well into the 1990s, thecommercially manufactured cumulative re-corder in laboratories devoted to the experi-mental analysis of behavior was as ubiquitousand essential as microscopes in a zoology lab.The importance of the cumulative recorderto the development of the experimental anal-ysis of behavior already has been noted. Formost of its history, in the experimental anal-ysis of behavior conducting an experimentwithout a cumulative recorder was like con-ducting an experiment without an operantconditioning chamber. In the 21st century,cumulative recorders are no longer manufac-tured and exist only in museums (amongSkinner’s bequeathals to the Smithsonian In-stitution’s National Museum of American His-tory in Washington, D. C. is an early RalphGerbrands Company commercially availablecumulative recorder) and the laboratories ofa few survivors from those halcyon days of op-erant conditioning.
Cumulative records, however, still have use-ful functions, certainly not as a primarymeans of data analysis, but as invaluable toolsfor monitoring moment-to-moment progressduring experimental sessions (cf. Ator, 1991)and for establishing standardization of exper-imental techniques (cf. Sidman, 1960, pp.343–347). For example, much contemporaryresearch in learning still uses schedules of re-inforcement as baselines against which the ef-
354 KENNON A. LATTAL
fects of different types of independent vari-ables are assessed. Thus having schedule controlover the baseline behavior is both importantand assumed prior to the introduction of theindependent variable(s). Because scheduleperformance is so well standardized (Ferster& Skinner, 1957), the cumulative record is es-sential in assessing whether the obtained re-sponding is indeed under schedule control interms of expected rates and patterns of re-sponding. Schedule performance differingfrom the schedule-typical performance for agiven schedule cannot be expected to yieldreliable, valid, and repeatable effects of theindependent variable(s).
The cumulative recorder is both a productand a reflection of the evolution of thescience of behavior toward greater quantifi-cation and precision. As with any human en-deavor, the recorder’s evolution, prolifera-tion, and ultimate demise are the result of thesame dynamic forces of variation and selec-tion that operate to shape and maintain otheraspects of our natural world. As the namesakeof the American Psychological Association’sDivision 25 newsletter, in some sense it liter-ally hailed the science of behavior. To thisday, a cumulative record appropriately ap-pears on the masthead of the Division 25 Re-corder. Students in introductory psychologycourses to this day as well still see cumulativerecords in their textbooks when basic learn-ing processes are discussed (e.g., Bernstein &Nash, 2002). The cumulative recorder alsohas come to symbolize both Skinner’s scienceand his personal characteristics of ingenuity,mechanical aptitude, persistence, precision indescription of data, and getting it right.
Ator, N. (1991). Subjects and instrumentation. In Iver-sen, I. H. & Lattal, K. A. (Eds.), Techniques in the be-havioral and neural sciences, Vol. 6: Experimental analysisof behavior (Part 1, pp. 1–62). Amsterdam: Elsevier.
Bernstein, D. A., & Nash, P. W. (2002). Essentials of psy-chology. Boston: Houghton Mifflin.
Bjork, D. W. (1993). B. F. Skinner: A life. New York: BasicBooks.
Carman, J. B. (1968). A template for estimating responserate. Journal of the Experimental Analysis of Behavior, 11,465.
Catania, A. C. (2002). The watershed years of 1958–1962in the Harvard Pigeon Lab. Journal of the ExperimentalAnalysis of Behavior, 77, 327–345.
Conference on the Experimental Analysis of BehaviorNotes (1948, June). Bloomington, IN: mimeograph.
Diamond, J. (1999). Guns, germs, and steel. New York: Nor-ton.
Dinsmoor, J. A. (1990). Academic roots: Columbia Uni-versity, 1943–1951. Journal of the Experimental Analysisof Behavior, 54, 129–149.
Dinsmoor, K. (1987). A special tribute to Ralph Ger-brands. Journal of the Experimental Analysis of Behavior,48, 513–514.
Emiley, G. S., Hutchinson, R. R., Hallin, E. R., & Kiraley,S. (1971). A convenient method for physical storageof cumulative records. Journal of the Experimental Anal-ysis of Behavior, 15, 248.
Ferster, C. B. (1953). The use of the free-operant in theanalysis of behavior. Psychological Bulletin, 50, 263–274.
Ferster, C. B. (1970). Schedules of reinforcement withSkinner. In P. B. Dews (Ed.) Festschrift for B. F. Skinner(pp. 37–46). New York: Irvington.
Ferster, C. B., & Skinner, B. F. (1957). Schedules of rein-forcement. New York: Appleton-Century-Crofts.
Frick, F. C., Schoenfeld, W. N., & Keller, F. S. (1948).Apparatus designed for introductory psychology atColumbia College. American Journal of Psychology, 61,409–414.
Gilbert, T. F. (1958). Fundamental dimensional proper-ties of the operant. Psychological Review, 65, 272–282.
Grindley, G. C. (1932). The formation of a simple habitin guinea-pigs. British Journal of Psychology, 23, 127–147.
Harrison, M., & Abelson, R. M. (1959). The maintenanceof behavior by the termination and onset of intensenoise. Journal of the Experimental Analysis of Behavior, 2,23–42.
Heron, W. T., & Skinner, B. F. (1939). An apparatus forthe study of animal behavior. Psychological Record, 3,166–176.
Herrick, R. M. (1965). Plotting and analyzing cumulativeresponse curves in operant conditioning studies. Jour-nal of the Experimental Analysis of Behavior, 8, 59–65.
Herrnstein, R. J. (1970). On the law of effect. Journal ofthe Experimental Analysis of Behavior, 13, 243–266.
Hurwitz, H. M. B. (1961). A bidirectional cumulative re-corder. Journal of the Experimental Analysis of Behavior,4, 331–332.
James, W. (1890). Principles of psychology. New York: Holt.Keller, F. S. (1975, April 29). Colloquium at Northeastern
University. Boston, MA.Keller, J. V. (1974). Modifying the Gerbrands cumulative
recorder to produce histograms. Journal of the Experi-mental Analysis of Behavior, 21, 381–382.
Killeen, P. (1985). Reflections on a cumulative record.The Behavior Analyst, 8, 177–183.
Kulli, J. C., & Bogrow, P. A. (1971). A cumulative recorderfor experiments on concurrent schedules. Journal ofthe Experimental Analysis of Behavior, 16, 381–383.
Lattal, K. A. (1973). Response-reinforcer dependenceand independence in multiple and mixed schedules.Journal of the Experimental Analysis of Behavior, 20, 265–271.
Lattal, K. A. (1999). Operant conditioning chamber. InA. Kazdin (Ed). Encyclopedia of Psychology. Oxford, Eng-land: Oxford Press.
Lekrone, R., Marmasse, C., Ferster, C. B., & Holtzman,E. (1963). An automatic paper folder. Journal of theExperimental Analysis of Behavior, 6, 269–270.
355HISTORY OF THE CUMULATIVE RECORDER
Lindsley, O. R. (1958). An easily copied ink for graphicrecorders. Journal of the Experimental Analysis of Behav-ior, 1, 14.
Michael, J. (1974). Statistical inference for individual or-ganism research: Mixed blessing or curse? Journal ofApplied Behavior Analysis, 4, 647–653.
Millner, E. C. (1970). A simple mechanical uptake rollerfor cumulative record data. Journal of the ExperimentalAnalysis of Behavior, 14, 199–200.
Morris, E., & Smith, N. G. (2004). On the origin andpreservation of Cumulative Record in its struggle for lifeas a favored term. Journal of the Experimental Analysis ofBehavior, 82, 357.
Morse, W. H., & Dews, P. B. (2002). Foreword to Schedulesof Reinforcement. Journal of the Experimental Analysis ofBehavior, 77, 313–317.
Parsonson, B. S., & Baer, D. M. (1986). The graphic anal-ysis of data. In A. Poling & W. Fuqua (Eds.), Researchmethods in applied behavior analysis (pp. 157–186). NewYork: Plenum.
Poling, A. (1979). The ubiquity of the cumulative record:A quote from Skinner and a frequency count. Journalof the Experimental Analysis of Behavior, 31, 126.
Robles, E., Alucema, F., & Cohen, J. (1981). A cumulativerecorder for two types of responses as a function oftime. Mexican Journal of Behavior Analysis, 7, 165–168.
Russell, I. S. (1961). Cumulative recorder ink. Journal ofthe Experimental Analysis of Behavior, 3, 24.
Sidman, M. (1960). Tactics of scientific research. New York:Basic Books.
Skinner, B. F. (1930). On the conditions of elicitation ofcertain eating reflexes. Proceedings of the National acad-emy of Science, 16, 433–438.
Skinner, B. F. (1932). Drive and reflex strength. Journalof General Psychology, 6, 22–37.
Skinner, B. F. (1933). The measurement of ‘‘spontane-ous’’ activity. Journal of General Psychology, 9, 3–23.
Skinner, B. F. (1938). The behavior of organisms. New York:Appleton-Century-Crofts.
Skinner, B. F. (1956). A case study in scientific method.American Psychologist, 11, 221–233.
Skinner, B. F. (1966). What is the experimental analysisof behavior? Journal of the Experimental Analysis of Be-havior, 9, 213–218.
Skinner, B. F. (1972). Are theories of learning necessary?In B. F. Skinner, Cumulative record (pp. 69–100). NewYork: Appleton-Century-Crofts. (Reprinted from Psy-chological Review, 1950, 57, 193–216).
Skinner, B. F. (1976a). Particulars of my life. New York:Knopf.
Skinner, B. F. (1976b). Farewell, my lovely. Journal of theExperimental Analysis of Behavior, 25, 218.
Skinner, B. F. (1979). The shaping of a behaviorist. NewYork: Knopf.
Snapper, A. G., Verhave, T., & Herman, R. L. (1964). Penfor cumulative recorder. Journal of the ExperimentalAnalysis of Behavior, 7, 422.
Street, W. R. (1960, March 1). March 1 in Psychology. Re-trieved July 26, 2004, from the American Psychologi-cal Association Historical Database Web site at CentralWashington University: http://www.cwu.edu/;warren/cal0301.html.
Todd, J. T. (2004, May). Did B. F. Skinner invent the cu-mulative recorder?: Pre-Skinnerian cumulative recording.Paper presented at the meeting of the Association forBehavior Analysis, Boston, MA.
Trudeau, M. (1990, July 27). Interview with B. F. Skinner.Washington DC: National Public Radio.
van Sommers, P. (1967). A cumulative recorder for stu-dent use. Journal of the Experimental Analysis of Behavior,10, 299–300.
Verhave, T. (1958). New ink for cumulative recorders.Journal of the Experimental Analysis of Behavior, 1, 56.
Wolkomir, R. (1996). They’re getting machines to pipedown. Smithsonian, 26, 56–65.
Yerkes, R. M., & Morgulis, S. (1909). Method of Pawlowin animal psychology. Psychological Bulletin, 6, 257–273.