THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER

By T. N. WHITEHEAD

HE human element in industry has been receiving increasing attention during recent years, and quite a number of researches bear witness to this fact. These researches, for the most part, fall into two classes. On the one hand, they are directed to establishing the fitness of particular individuals for given jobs. Tests of intelligence and trade tests come under this head. On the other hand, a large number of investigations are undertaken to improve the traditional work procedures, and thus enable operators to achieve a greater result with a smaller expenditure of energy. The work of the Gilbreths occurs to one in this connection. Both these types of research have proved fruitful in the past and will undoubtedly extend their scope in the future, but neither of these lines of investigation, separately or collectively, amount to a study of the worker himself. For behind his individual facilities, and the orderly arrangement of his movements, stands the human being; something more than a collection of physiological possibilities and coordinated actions. So the study of the human element in industry must include the whole gamut of hopes, desires, fears, and interests, in so far as these motivate and organize work behavior in its widest sense. The Human Element in Industry Hawthorne Experiment In 1927, at the Hawthorne Works of the Western Electric Company

in Chicago, a research was started to attack this very problem. Perhaps it would be more correct to speak of a number of related researches, for the program continually widened over several years. In this account I shall confine myself to one experiment which lasted for about five years. I should explain that the research I am describing is not completed; the analysis is proceeding now. In 1927, five skilled operators, young women, were placed in a special test room and performed their usual work under continuous observation until the middle of 1932, a period of five years. These young women all had had several years' previous experience in the assembly of small electrical relays, and this was the work they performed in the test room. These relays weigh a few ounces, contain some 40 to 50 parts, and are assembled in an average time of just under a minute. Although the actual work itself was not altered in any significant manner, nevertheless the general conditions under which these operators worked in the test room differed in a number of respects from those to which they had previously been accustomed. The operators were informed as to the nature of the experiment. They were paid, as before, on the same system of group piece work; but the group, formerly containing over a hundred individuals, was now reduced to only five. In addition, no pressure was put on this group to achieve any given level of output; and, in fact, they were

458

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER warned against racing or forced output in any form; for the immediate object was not to increase output but to study the human factor. The operators were seated in a row on one side of a long bench, and were permitted to talk. Conversation was the rule rather than the exception; it was sometimes general, and at other times was confined to more intimate conversation between neighbors. In addition to these innovations certain other experimental changes were introduced from time to time. Thus the length of the working day and the number of working days per week were varied; rest pauses were introduced, and so on. Provision was made to keep adequate records. One or more executives were stationed continuously in the test room throughout the whole experiment. They had certain routine duties to perform in connection with the supply of parts and the disposal of the assembled relays. In addition these executives, with the full knowledge of the operators, kept an elaborate series of records and observations. Thus, an automatic instrument recorded, to about a fifth of a second, the instant at which each operative completed every assembled relay. So we have a minute-to-minute record of output for each operator over a period of five years. Other records relate to the quality of the output, and also to the quality of the parts supplied; reasons for temporary stops in the work; records of conversations; room temperature and humidity, etc. Besides these more formal records the executives made observations on the characters and dispositions of the workers in the test room, and recorded the relationships that developed between them. I should perhaps add

459

that the relations between the operators and the executives were at all times cordial. These operators were separately interviewed by an experienced interviewer on several occasions in another room; and finally they were examined by a medical officer about once in six weeks. Weekly Rate of Output Figure I shows one way in which the output records can be presented. This diagram gives the weekly rate of output of every operator in the test room throughout the whole duration of the experiment. The number of each operator corresponds to the number of the seat she occupied, counted continuously along the work bench. It will be seen that occasionally one operator left and another took her place; thus, at about the end of 1927 operators ia and 2a were replaced by operators i and 2; again for about ten months in 1929 and 1930 operator 5a substituted for operator 5. In about the middle of 1932 all the operators left and were replaced by novices. The present account concerns only the five operators i, 2, 3, 4, and 5. From April, 1930, to February, 1931, the operators changed places on the work bench and occupied the positions indicated; they subsequently returned to their old positions. These weekly rate-of-output graphs present only one of many possible summaries of the operators' work behavior from 1927 to 1932; but even this single summary cries aloud for interpretation. For instance, why do a number of highly skilled operators, whose outputs had already reached a steady state, work so much faster after entering the test room ? Why do the working rates

46o

HARVARD BUSINESS REVIEW

Chair N3

80 70 60 50 1
30 20 10 80 7D

. ,. . <}prafor

S Jilt.

w
A
1

11
H
Chair h4a4 1
1

1 *1 1

tt t t t t

t It

t t

-0

I h
Of. 4

)
-

Jr n
1
+-^
1**1 1 ^

f^
Chair Nal[

Chair Na4

Chair

50 40 30 30 ft t t t t 70 i 60 50 " 30 1

t t*^ t t

- ,

IT UL HE

k
1

r
1
J5Sk Op.Sii .kl

S^
1

H
Chair - N B S 1

1
1
1

1
l3'
7'8*1

jj

11* 1 * 2

1 3 $

1 7

* IIS 8

M a..yrTTTTa O 2li

nding Sa'Kirolay
1927 I ms I IW9 I 1930 I 1931 1 1932 I

FIGURE I.

WEEKLY RATE OF OUTPUTAVERAGE NUMBER OF RELAYS ASSEMBLED PER HOUR

(Relay Assembly Test RoomHawthorne Works)

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER of numbers 2 and 4 continue to rise more abruptly than the rest, and then suddenly drop in 1930? And finally why does No. 4 become so singularly irregular in 1931.^ These and many other questions are prompted by this diagram. The data collected during the course of this experiment can be analyzed from several points of view. In the first place, we get a very interesting insight into the technique of this particular repetitive work. Thus, by analyzing the relay-byrelay speeds of assembly for any one operator, it can be shown that the number and the extent of her deviations from her own mean speed obey a definite law. These deviations are just those that might be expected to arise from the thousand and one small causes we call "chance"; they do not arise from the influence of either one of two major disturbing causes. Moreover, these chance variations refer, for the most part, to the operators themselves, rather than to any variation in the work. Figure 2 shows a frequency curve for operator No. 2, and the shape of this curve, a close approach to a logprobability curve, is evidence of the "chance" nature of the scatter in speed of assembly. But this curve takes no account of the order in which these various speeds of assembly occur. It is possible that, although the frequencies of the various speeds are random, they may, nevertheless, occur in some particular order. The actual order in which the various speeds of assembly occur has been examined for a number of separate time spans of an hour or so; and in the case of each operator it is found that, apart from an occasional slow trend, the arrangement of fast and slow

461

assemblies is just what might be expected from chance. Again, it is interesting to find that when an operator takes an abnormally long time to assemble a relay, owing perhaps to some blunder or accident, this in no way affects the probable speed of her next assembly. Exactly the same thing can be said of exceptionally fast assemblies. All this presents a picture of operators working well within themselves
Operotdr Not-Bilay Typt R OK 6

1 =
i>
FiClTKE 2 .

ndayt in June, July, Au9,Nn

>.,

io 1) Seconds
FKEQUEHCT

60
CUKVB.

and in accordance with their natural gait or swing. The effect of working above or below a natural rhythm is to upset the random frequency of variations, and also to upset the random ordering of these frequencies. Again, it can be shown that the actual rate of physical movement of any given operator is about the same whether she completes a relay in 40 or in 80 seconds; that is, whether she works fast or slowly. Thus, in 1930 the operators were assembling relays faster than in 1927, but their physical movements were about equally rapid on the two occasions. Fast work occurs when the operator has acted with great skill, and thereby has diminished the number of small discriminative finger movements. In the case of slow work the number of discriminative actions has been in-

462

HARVARD BUSINESS REVIEW Here is a list of the physical factors considered: (1) Changes in the length of the working day. (2) Changes in the number of working days in the week. (3) Working days with and without rest pauses. (4) Frequency of changes in type of relay to be assembled. (5) Changes in the quality of the parts supplied(6) The number of hours during which the operators were in bed on the previous night. (7) The effect of vacations and holidays, both in anticipation and in retrospect. (8) The periodic illnesses of the workers. (9) The daily, weekly, and annual cycles. (10) Changes of temperature and of relative humidity. We have very adequate data on all these physical factors, in many cases stretching over the whole five years. And we have examined these factors to see whether they account for changes in output rate. The answer appears to be much the same in every single case. The output was never affected much, and often not at all, by any change we can find in physical circumstance. This is rather a surprising result, and I will give some of the findings in more detail. Hours of Work.The hours of daily work, and the number of working days in the week, were varied on several occasions. Each arrangement lasted for a number of weeks, and often for several months. The hours worked per week varied from 48 down to about 30- The number of days worked per week varied from 6 to 4.

creased by their inferior adaptation to the exact circumstances of the case. In this particular occupation, rate of output seems to be connected with the operators' attention, rather than with their actual speed of movement. It was the increased quality of their attention which enabled these already skilled operators to increase their rate of output after entering the test room. But plenty of evidence shows that this improved work must be ascribed to some general change of attitude, some heightening of their work interest, rather than to anything in the nature of increased determination, or will power. The fact is, although these operators were interested in their output, they were never forcing themselves over the course, above their natural inclinations- What changed was precisely the measure of their natural inclinations. Physical Circumstances Affecting Work Now what were the main factors influencing the work behavior of this group t We saw in the first figure that the output rate of these operators was subject to a number of changes, sometimes downwards, but mainly up. If changes in the circumstances of these operators can be shown to be coinciding, or correlating, with changes in their work behavior, then it is a fair presumption that these circumstances were important to their motivation. Conversely, if changes in some particular circumstance are not found to correlate with changes of output rate, then, within limits, this circumstance is not a factor in motivationFirst, a number of physical circumstances were carefully examined to determine how far their variation affected output rate.

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER On only one single occasion could the slightest influence on the output rate be seen. Once, when the hours were put to the maximum and all rest pauses were eliminated, a downward dent appears in the general trend of the curves, September and October, 1928. The output rate, even then, showed no signs of descending to the groups' original rate after entering the test room, although the conditions were identical on the two occasions. Apart from this, these changes did not infiuence output rate. Changes in Type of Relay.Now let us consider the frequency of changes in type of relay. I have been writing as though the operators assembled only one type of relay throughout the whole experiment. But this is not so. Each operator assembled over 150 different types of relay between 1927 and 1932. Some of these types are almost indistinguishable from each other; others differ widely. The diagram of output rate, as was shown earlier, was arranged to compensate for these differences. All output has been converted to a standard type, and so the varying difficulty of the different relays does not affect the character of the output diagrams. Nevertheless, it is entirely possible that the fact of changing from one type to another should affect output rate. The number of times that any given operator changed relay type per week is a very variable figure. Sometimes they continued the same type for weeks together, whilst at other times they may change type 5 or 10 times week after week. For two years No. 5 averaged something like 20 changes of type per week, the weekly figure varying from under 10 to over 40. Nevertheless, the variation in the frequency of these changes completely

463

fails to correlate with changes in output rate. Another study was made to see whether the output rate during the first hour, immediately after changing type, differed from the output rate on the same type during the succeeding day. This test could only be made on types that ran for at least two days, but this was quite a common event. The answer is that the output rate, during the first hour of a changed type, was not distinguishable from the rate on the next day. So no disturbance due to change of type can be detected, in spite of the fact that each type demands its own particular pattern of behavior. The parts are different, and are assembled in a new order, and in different quantities. This is a surprising result but is well authenticated. Vacations and Holidays.What effect do these have on output rate ? The rate of work tends to be a trifle lower during Christmas week, perhaps I % or 2% on the average. This is not always noticeable; and neither the single day holidays nor the two weeks' summer vacation have the slightest effect on output rate. Daily and Weekly Cycles.The story is much the same whatever change in physical circumstance we consider. For instance, take some of the cyclical changes of circumstance. The daily cycle is a good example. Figure 3 shows a cumulative output curve for operator No. 2 for a given day. The working day is divided into four nearly equal parts by two rest pauses, one from 9:30 to 9:45 A. M. and the other from 2:30 to 2:40 P. M., and finally by the lunch interval, 12:00 noon to 12:4s P- M. Each quarter of the day has been plotted from a common base to save paper, and the

464

HARVARD BUSINESS REVIEW

method of plotting is as follows: every characteristic of this diagram is the dot corresponds to one relay assembled comparative absence of the influence and is plotted a standard distance of the daily cycle on output rate. This above the dot below it. Thus, the is quite typical. vertical distances between successive The same sort of thing applies to the dots are all equal. The horizontal weekly cycle. Typically, the first and distances between successive dots are last days of the week carry the lowest variable and depend on the time taken output rates; and this applies whether to make the various relays. The hori- a six-, five-, or a four-day week is
I V 12:00 n 120
/ ft

&00p /

100

/ y
/ 20

/ /
12:45 p

y
0

iy

7:30a

9:45ci

/ Z40p 1300 3^000

FIGURE 3.

4^500 6^000 '?500 ^000 Seconols CUMULATIVE OUTPUT CURVE FOR OPERATOR N O . 2 FOR JULY 61928 (FRIDAY)

zontal time scale, in seconds, is given at the bottom of the figure. Thus, if all the relays take about the same aver^ age time to assemble throughout a quarter day, then the line will be substantially straight; and the line will be more nearly vertical the faster is the average rate of work. This is a very typical working day. Perhaps there is a slight warming up period at the beginning of the day; it is very slight. The most noticeable feature is a comparatively poor output for the first half of the afternoon, with a complete recovery after the 2:30 rest pause. There is nothing corresponding to an end spurt; and the outstanding

being worked. But the difference in rate is very small and is by no means an invariable feature. Again no seasonal fluctuation in output rate can be found. Temperature and Humidity.-The last physical factors I shall consider are temperature and humidity. First, the room temperature and humidity have been analyzed in a number of different ways. Thus, the average temperature for the working hours entirely fails to correlate with the amount by which the output rate for that day stands above or below the average rate for the period. The same applies to relative humidity.

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER This applies not only to temperature and humidity, but also to the two taken in combination. Thus, output was not affected by the simultaneous event of a very hot and a very humid day. If we examine the data, not on a dayby-day basis, but on a weekly average, the same story holds good. Spells of humid, or of hot, weather have no more effect on output rate than do single hot or damp days. The analysis of the outside, as distinct from the room, temperatures and humidities leads to exactly the same result. Temperature and humidity, whether in the room or outside, have no effect on output rate in the case of this group. Slight Effect of Physical Factors.I cannot devote any more time to describing the effect of changes in physical factors on output rate. But the findings are uniformly negative, or at least very slight. And this again leaves us with the picture of a group working well within their physiological limits; with a margin, or reserve, from which they can draw to meet the common run of difficulties and daily ups and downs. It is in virtue of this physiological reserve that they compensate for these very significant changes in their physical circumstance. Of course, the work is of a light repetitive character, but we must not forget that thesi.e particular operators reached a level of output which easily exceeded that of the regular shop. These operators compensated for changes in physical circumstance, such, for instance, as a changed relay type; but not with the result of holding their output rate constant. On the contrary, as the first figure shows, their rate of work was continually changing, both week by week and in longer trends.

46s

Nevertheless, this ceaseless change in output rate is not dictated by change of physical circumstance. We face two major problems. Why are the precise details of physical circumstance so unimportant to these operators ? And, what are the important elements in the motivation of this group ? Perhaps these two questions will admit of a common answer. Work of Two Operators Compared If we glance back at Figure i and look at the graphs for operators 2 and 4, certain similarities will become apparent. In both cases the rate of work rises unevenly until about December, 1930, when, in the case of both operators, the output suddenly drops. No. 2 suffers only a moderate drop in speed and continues on the same average level for the next twelve months or so. No. 4 drops more dramatically and continues to fall slowly for the next year; moreover, the irregularity of her weekly graph becomes very pronounced throughout this latter period. Figure 4 contrasts these two operators in another way. In the middle graph the output rate graphs for operators 2 and 4 have been copied from Figure i without any alteration. When the line for 2 lies above that for 4, the space between the lines has been filled up; in the reverse case it is left blank. In this way the comparative rates of work of these two operators can be easily compared. The difference between their rates is plotted in the top graph for greater convenience. You will see from this diagram that No. 2 is a slightly faster worker than No. 4, but the difference, for the most part, is not very great. On occasion their speeds became almost identical.

466

HARVARD BUSINESS REVIEW

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER This is shown very clearly in the top graph. Now look at the bottom graph; this shows the degree of correlation existing between the daily output rates of the two operators from time to time. The scale on the left hand side is called the "Coefficient of Determination." "Determination" is the technical name given to a particular method of calculating correlations; it has advantages which need not concern us here. You will notice that near the end of 1930 the correlation between the daily output rates of the two operators is nearly perfect. This does not imply that their rates are identical, though this happens to be the case. What a perfect correlation does mean is that the rates of the two operators are moving together, so that if one rises the other will rise in the same proportion, and vice versa; they rise and fall together. It is important to the argument to remember that these correlations are telling us nothing about the absolute speeds of the two operators, which may be very different. Nevertheless, in this particular instance, whenever the speeds of the two operators tend to coincide, then the correlation between their variations of speeds also rises, and vice versa. There is nothing in the mathematical procedure to make this probable, and it does not occur between any other pair of operators. Characters and Social Relationships of Workers To interpret this we must know something about the characters of the operators and their social relationships. No. 2 was very definitely the leader of the group, partly because she was the fastest worker, but also in virtue of her character. Of Italian extraction, and in her late twenties. No. 2 is the

467

ablest of the group. She possesses, in a high degree, those personal qualities which mark out a woman as a social leader: quick-witted, ambitious, decisive, a sense of responsibility, and a ready temper. Owing to the death of her mother. No. 2, in effect, ran her father's household; and, by her earnings, materially contributed to the upkeep and education of her younger brothers. Operator No. 4, of Polish origin, is one of a fairly large family, and she lived with her parents in a Polish district in Chicago. The family are hard workers, frugal and self-respecting. The seven children received a more than average good education, one brother graduating from college. No. 4 is intelligent, independent, even tempered, and hard working. She formed a friendship with No. 2 in the early days of this experiment, and this friendship continued unbroken throughout these five years. Without having outstanding capacities of personal leadership. No. 4 was normally ambitious, and at times undoubtedly regarded herself as a rival to No. 2 as a group leader. Two very different social attitudes are illustrated by the relationships of the various operators. On the one hand, there is the friendship, or interest, which one operator had in another as an individual. This is paralleled by the very usual friendships between two persons anywhere. This sort of interest involved only two operators, usually sitting next to each other, and was not related to the group in any marked degree. Another social attitude involved the relations of an operator to the group as a whole. Thus, No. 3 was in some degree the "wag" of the party; her general conversation was directed to

468

HARVARD BUSINESS REVIEW both their Monday output rates are 15% to 20% below the average rate of the succeeding days. It is as though both operators agreed to call the race off every Monday, although I think the mechanism is more unconscious. This symptom was not marked in the struggle of 1929. Finally, No. 4 again accepts defeat and her rate broke dramatically. No. 2 also drops her rate, but not so much, and she regains a substantial lead over No. 4. Simultaneously, the correlation of daily rates of output sinks to zero, and the abnormally low Monday rate disappears. Shortly afterwards the operators return to their old seats. The next year is one of undisputed leadership for No. 2, and her output rate remains substantially constant over this period. No. 4, on the contrary, shows sonie very characteristic symptoms; her output rate continues to fall slowly and, in addition, her weekly output becomes unusually irregular. Some work done by Elton Mayo, when investigating a cotton mill, and another experiment at Hawthorne, both lead to the conclusion that abnormal irregularity is often produced by what may bie described as "unpleasantly tonied preoccupations," a condition of "semicompulsive reverie," "brooding." At the beginning of 1932, for a brief period No. 4 makes one last effort to lead the group, and the same symptoms reappear. The relations of 2 with 4 form a very complete contrast to those of 2 and i. No. I is a fast worker and has a pleasant relationship with her neighbor. No. 2. On occasions No. i works practically as fast as No. 2. On other occasions the daily fluctuations of output rate of these two operators show a high correlation. But, and this is the

producing a laugh, and sometimes at her expense. She featured herself in droll and unusual situations. No. 2 was related to the group as its leader, a position admirably suited to her temperament and aspirations. No. 4 was related to No. 2 by mutual personal friendship. But she. No. 4, also saw in herself a possible leader of the group, and so had a group interest, tn this capacity she found herself as a rival of No. 2. Here was a personal relation between 2 and 4 which depended for its existence on a mutual group consciousness, in which both operators desired the same unique position. The result we see in these diagrams. Although 2 and 4 were not sitting together. No. 4 makes a bid for leadership in 1929. This she does by trying to exceed the rate of work set by No. 2; No. 2 keeps just ahead, step by step. You will also notice that their daily fluctuations of output rate correlate to a very significant degree. At the end of the year No. 4 abandons the struggle, or at least slackens in her efforts, and falls below No. 2. The correlation between their daily output also sipks, and this rivalry might perhaps have faded away. But early in I93(^ for experimental reasons, the operators were all allotted different seatsy and No. 4 finds herself sitting ntixt to No. 2. No. 4 is again conscious of her status as subordinate to No. 2, and she makes another effort to capture the leading position. The result is plain; 4 steadily increases her rate; and z, by a very narrow margin, maintains leadership. Again their output rates step up in almost perfect synchronism, until towards the end of 1930 they are both working at record speed. But a note of strain begins to creep into the picture. Throughout these last weeks

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER point, the periods when their output rate8 fluctuate in unison are not the same periods as those in which their speeds are nearly identical. No. I always accepts No. 2 as her leader and is happy in this situation. She often follows No. 2 in her daily fluctuations quite irrespective of whether their outputs are nearly equal or not. . In a general way the curve for No. i is a sort of imitation of that for No. 2. For No. I is more dependent on No. 2 than vice versa. It can be shown that in a number of ways their work behavior varies together. To a less extent No. 3 has the same relationship to No. 4, as No. i has to No. 2. No. 3 is no rival and to a great extent leads an independent existence, but her behavior shows unmistakably the influence of No. 4. Thus, she alone slightly decreases her rate in 1931, in sympathy with the more pronounced decline of No. 4. In the case of both i and 3, their correlations with their respective friends' outputs practically ceased when they were separated in 1930. These correlations revive when the operators return to their old seats in 1931. Space does not enable me to develop a number of other ways in which these operators affect each other's outputs, but this mutual influence is very far reaching, and shows itself in a surprising variety of forms. Menial Attitude and Human Relationships Two major generalizations apply to the behavior of this group of workers from 1927 to 1932. In the first place, significant changes in work behavior can almost always be traced to changes in mental attitude.

469

rather than to changes in physical circumstance. Secondly, these changes of mental attitude usually, though not always, relate themselves to the workers' human relationships, either as between themselves, or as between the workers and the management. And this suggests a rather obvious, but very important, truth; namely, where operators are working well within their physiological possibilities, their work behavior will be conditioned by their work morale. That is, their behavior will depend on the force and the liveliness with which the work situation captures their imagination, and tones their thoughts. I have tried to show you that the element of human relationship is a very large part of the work situation to the worker himself. Financial Incentive The financial incentive ought to be given its place in the scheme of motives. If the firm had not paid wages, no relays would have been assembled. Moreover, it is well known that, of various alternative payment systems, some encourage better work than others. But even here, the occasional failure of a well-tried system of payment warns every experienced executive that wages are a necessary, but not a sufficient, condition of good work. Let us turn once again to Figure i. All these operators are paid on the same system of group piece work. But, of course, their pay has not the same significance for the different individuals. I have told you something about the home life of No. 2. She earned more than her father, the only other wage earner in the family, and was the main support of the home. This operator accepted the double role of foster

47O

HARVARD BUSINESS REVIEW jobs, or accepting reductions in pay. The financial incentive rises, but no corresponding rise in output rate occurs. Finally, the situation becomes acute about May, 1932. Several of the operators are in financial straitsNo. 2 is struggling, against a difficult financial situation, to hold her home together. To cap all, the group learns that, as young employees with only a few years' service, their turn has come to leave the firm. The axe has at last reached them, and they are given a few weeks' notice; but when they leave they will be paid a lump sum over and above their wages. As always, the firm is doing well by its workers; and nothing is more revealing than the steady belief in the humanity of the firm's policies, on the part of all its employees. Nevertheless, something not far short of semistarvation, or charity, faces several of this group. But a few weeks' employment are still left; and a determined effort to raise output rate would do something to ease their situation. The actual fact is that output rate fell violently throughout these last few weeks. The financial incentive has failed exactly when it is most compelling. Like the rest of us, the employee has a logical reason for her actions; she consents to assemble relays day after day in return for pay; and, since the pay is graded according to the work accomplished, she desires to work faster and earn more. This is the logical thread which guides and organizes the workers' total situation, but it is not the whole story. We all know the difficulty of a long continued effort against the grain; and long hauls are only accomplished in virtue of an immediate sense of the importance of the experience. Thus

mother and chief wage earner for the younger family, and her wages provided the material framework of this situation. Consequently, we find that her conversation with the group is continually reverting to the desirability of achieving high outputs. Her own output starts on a high level and rises until near the end of 1930. She maintains a high output from then on. Moreover, No. 2 influences others in the same direction, notably her friends Nos. I and 4. At the opposite end of the scale stands No. 3. Of Polish extraction, she entered the test room at the approximate age of 21, and lived at home with a taciturn father, a dominating mother, and a large family of brothers and sisters. The family was impecunious and the atmosphere was one of hardship. No. 3 handed her weekly pay check over to her mother and was given a very small weekly allowance. Moreover, since she held no responsible place in the family councils, the expenditure of her money had no significance for No. 3. It is as though No. 3 came to work, not for pay which she never spent, but because her mother sent her. The financial incentive was largely lacking. You will notice that her output rate ceases to rise much after the middle of 1928, and pursues a nearly even course from then until the end. In this contrast we see the function of the financial incentive working according to theory.
Effect of Depression on Output

During the last year of this group's work in the test room the depression was at its height, and the whole firm was working short hours. At the same time other wage earners in these operators' families were losing their

THE SCIENTIFIC STUDY OF THE INDUSTRIAL WORKER No. 2 needs the money but finds her motivation in a race with No. 4. In the end the operators face real poverty, but the social situation is evaporating; it has no meaning and output goes down, As No. 2 explained to me, "We lost our pride."

471

Once again action is organized on a logical basis, but this does not provide its motivation. Work is finally performed because here and now the situation satisfies the need for selfexpression; because it provides an acceptable way of life.

Harvard Business Review Notice of Use Restrictions, May 2009 Harvard Business Review and Harvard Business Publishing Newsletter content on EBSCOhost is licensed for the private individual use of authorized EBSCOhost users. It is not intended for use as assigned course material in academic institutions nor as corporate learning or training materials in businesses. Academic licensees may not use this content in electronic reserves, electronic course packs, persistent linking from syllabi or by any other means of incorporating the content into course resources. Business licensees may not host this content on learning management systems or use persistent linking or other means to incorporate the content into learning management systems. Harvard Business Publishing will be pleased to grant permission to make this content available through such means. For rates and permission, contact [email protected].