Hollerith Punch Card
The Hollerith Punch Card system has been widely used since the late 19th Century as a means for data-storage and tabulation. It was created out of a need to speed up the 1890 census at a time when machines were rapidly replacing manual labor. The need for the Hollerith system seemed pressing because the U.S. government required more information from the population, and therefore more advanced techniques for tabulation required for statistical analysis. The Punch Card technology was a remediation of card catalogues, short hand writing systems, and the Jacquard Loom. Although in recent years the Punch Cards have fallen out of favor, traces of the original cards are still seen today. The influence of the Punch Card system can be seen in the computer, bureaucratic record keeping, scan-trons, and voting machines.
- 1 The Precursors to the Punch Cards
- 2 Socio-Historical Context
- 3 Initial Design
- 4 Multiple Uses of the Punch Card Technology
- 5 From Punch Cards to Magnetic Tape
- 6 References
The Precursors to the Punch Cards
The Hollerith Punch Card system was formed out of a need for a more accurate, rapid census and as a labor saving device. The idea for the punch cards had first been introduced in the 18th century by Jacques de Vaucanson, who invented a card-controlled automatic loom. Here patterns were coded on a sequence of cards, and were then said patterns were executed on silk. (George Dyson, "The little secret that haunts corporate America ... A technology that won't go away." Wired magazine, Issue 7.03, 1999) In 1804, the card technology was used by Jacques Jacquard as an improvement on the Jacquard loom. These cards were used to control needles to produce the woven product. The desired pattern was encoded on cards to be reproduced. Charles Babbage, in 1848, envisioned a Punch Card system, but did not have the technological knowledge for the invention. The lithograph, although not a direct precursor, was a technology that also reproduced visual patterns in the 19th century, with the prospect of making the original visually reproducible. This shows a trend of the shift towards mechanical reproduction, and the use of machines to facilitate processes. Another device that used a similar logic to the Hollerith system was the Piano Roll. Although there is no evidence that shows a connection between the Punch Card system and the Piano roll, there is a similar logic that exists between the linearity of the Piano Roll, and the nascent version of the Punch Card roll.
Another direct precursor to the Punch Card system was the Automatic Telegraph. In the patent for the Automatic Telegraph, the description is as follows: "the transmission is effected by the means of an automatic controller consisting of a pattern strip having predetermined characteristics...which characters are telegraphically reproduced at the receiving station." (Patent 255332) This mimics the punch card system in its representation of information by the use of patterns. Another form of the Automatic Telegraph utilized the idea of perforating paper, that was rolled over an electronic drum. Wherever the holes appeared, a connection was made, and a signal was sent directly to the telegraph. (Patent 561547) These inventions paved the way for the technology used in the design of Hollerith's Punch Card system.
The historical context of the Punch Card system shows a paradigm shift in the proliferation of knowledge pertaining to the general population, and the shift in the use of human to machine labor. Technological innovations of the time, showed an increase in the use of machines that were intended to encode information. Machines were created to streamline processes, as well as improve accuracy. The sentiment regarding the use of machines is shown in this quote taken from a journalist in 1890, in his description of the Punch Cards "(Hollerith) simply made it possible to do a certain work with astonishing rapidity and absolute infallibility." (Headline: A Boon to Tabulators Doing the Work of the Census Department by the Hollerith System; Article Type: News/Opinion Paper: Inter Ocean, published as The Weekly Inter Ocean; Date: 04-15-1890; Volume: XVII; Issue: 4; Page: ; Location: Chicago, Illinois) The work that this journalist was referring to was the American census.
The role of the Punch Cards in the American Census
The first American Census took place in 1790 and was carried out through the sheriffs and deputies of the colonial governments. The said purpose of the census was to provide a “thorough enumeration of the population,” to count the number of people for tax and voting purposes( How A Census is Taken-The North American Review- Carrol Wright U.S. Commisioner of Labor (1889)). The census, which was pursued knowledge of the nation's population, was done in order for governments to keep track of the number of people that existed, and was not originally intended for knowledge of detailed statistics. The first census only required a minimal amount of information compared to the 1870 census. It did not require the exact age of the person, rather whether or not said person was above or below the age of 16. White men were counted as people in this respect, while they age range of women was unnecessary. The reason for this was that women could not vote at the time. The number of families were counted, and finally the number of slaves. There were only 6 boxes that had to be filled in at this point. Carroll Wright, the U.S. Commissioner of Labor believed that the results of the census were faulty, as there was a large margin for human error.
The census in 1880 was vastly more detailed then then its predecessors. It required a more intricate and intimate knowledge of facts about the relationship of members in a household, their mental condition, (insane, idiotic), the products that they bought, etc. Under the Garfield Bill, the organization of the census changed such that there was now a Superintendent of the Census with an agency of stenographers, statistical experts, clerks etc that consisted of “at least 1500” persons. The cities were divided up into districts, where a large number of people were hired, so that there was a better ratio of enumerators households. At the time there were 35,000 enumerators working for the government. These people collected the facts by going from door to door and hand writing the data. The enumerator would count all of the people, and then would forward the numbers to the superintendent. At this point, data collected for the census was hand tabulated information marked on cards, and then sorted by hand in various ways. The cards were subsequently counted by hand. The method of writing on cards was then replaced by the Hollerith system which automated the census. Hollerith’s Art of Compiling Statistics: (1989) (Patent 395782) Idea for automation of the census. Machines could not read hand writing, but could read cards with holes in them. The system seemed necessary because it could count combined facts, collate the data, and speed up tabulation for the 1890 census.
Hollerith in the Age of Biopolitics
Alexander Galloway ventures to say that the birth of biopolitics in the United States could very well be pinpointed to the 1890 census, which was the first to implement the tabulation system of Herman Hollerith (Galloway, 80). To be sure, the tabulation of information about individuals, not simply a count, was a move away from the romantic aura of the individual as free-standing, stable agent. As Walter Benjamin suggested about art some decades after 1890, the individual as an ideological remnant of Enlightenment, would soon be reconstituted by the statistical tabulation of its supposed properties—race, health, sex, etc. Indeed, in the years leading up to the turn of the 20th century, in which modernity would crystallize into its late, “full blown” form, the socio-historical “diagram” of biopolitics that Foucault outlines in Society Must Be Defended and the first volume of History of Sexuality (among other places) can be readily discerned.
As has already been commented upon by many scholars, Foucault coins the term to identify the process by which governance begins to work below the level of the individual (i.e., subjectivity) to the level of life (i.e., the biological). More specifically, he argues that the 18th century is when the problem of technical knowledge emerged. The State launched a political economic struggle aimed at reigning in various subordinated knowledges of this period, in order to select, normalize, hierarchicalize, and centralize it under the banner of a dominant (read: State), yet multiplicitous knowledge (Foucault, 180).
(1) Selection: The elimination and disqualification of economically expensive knowledges.
(2) Normalization: Making various knowledges communicate with one another and to break down the geographical and technological boundaries between them, effectively making them interchangeable.
(3) Hierarchicalization: The hierarchical classification of knowledges, thus interlocking them.
(4) Centralization: Centralization allows for knowledges to be controlled pyramidally, that is, knowledges can be transmitted upward from the bottom, and also downward from the top.
How this happens, Foucault suggests, is through the production of populations through databasing, one in which the individual is transformed into a “dividual,” (Deleuze) signaling a new era of power, in which the masses are made into statistics, what in today’s world of the health care industry, for example, with its attentive administrative record-keeping would be considered a cruder, less sophisticated version of bioinformatics. As he says, “The disciplinarization of knowledges, and its polymorphous singularity, now leads to the emergence of a phenomenon and a constraint that is now an integral part of our society. We call it ‘science’(Foucault, 182).” In Hollerith’s system, each card represents one person, with the various punch holes representing “some particular value or meaning is assigned; a hole in one place meaning, [for example,] a white person, in another black. . . In this way, we not only recorded the answers to the twenty-six inquiries of the population schedule, but we also recorded the particular [s]tate, county, city, and enumeration-district in which the given person resided” (Hollerith, 679). As Foucault describes, the individual in the age of biopolitics is neither the individual of social contract theory nor that of the theory of rights but something quite different, inextricably tied to a “population.” “What we are dealing with in this new technology of power is not exactly society (or at least not the social body, as defined by the jurists),” Foucault writes, “nor is it the individual-as-body. It is a new body, a multiple body, a body with so many heads that, while they might not be infinite in number, cannot necessarily be counted. Biopolitics deals with the population, with the population as political problem, as a problem that is at once scientific and political, as a biological problem and as power’s problem (Foucault, 245).”
The Problem of Demographic (Read: Politico-scientific) Identification in the late-19th to early-20th century
The problem of social identification was an issue that came to a boiling point during Reconstruction, after the so-called emancipation of enslaved descendants of Africans at the end of the Civil War (1861-1865). If there was any doubt before, race had become the primary “trope” (Henry Louis Gates 1993) of American life. Though slavery would be outlawed, the equally pernicious system of Jim Crow would maintain racial separation as one of the primary functions of the American political apparatus. But even prior, in the 18th century, when the institution of slavery was not debated, the status of enslaved Africans in America was a sticking point of discussion for reasons pertaining to political representation. The 3/5 clause of 1789, for example, counted slaves as 3/5 of a human (that is, a white) in determining the allotment of representatives per congressional district, not to mention the well-known “one drop rule” that pronounced anyone with any African ancestry to be considered “black,” that is, as slave. Hence, the states that would later make up the Confederacy would hold greater congressional power, based on their slave population. Moreover, it is worth noting that the racist fractional consideration of enslaved Africans did provoke a necessity of a governmental tabulation of populations in order to calculate congressional representation. Thus, Foucault’s assertion that the State would be interested in accumulating a dominant, technical knowledge would be latent until the demographic revolution, for which Hollerith’s Electrical Tabulating System would be responsible. The political quanta of citizenship would at this moment be elided with one’s status as “human.” The political and biological would from here on be interimplicated ontologically.
By the period of Reconstruction, it became necessary for the State to keep track of its populations as the Union struggled to heal its wounds while tarrying with the fundamental contradiction of Jim Crow. Black Americans were no longer slaves (hence, theoretically citizens), yet not fully human in the eyes of the law (and the majority of whites) since they were not yet given suffrage. So the maintaining of Jim Crow and later de jure segregation during Reconstruction, as W.E.B. Du Bois suggests, was a matter of reconstituting identities and identifications into fundamentally mutually exclusive categories, like that of black (not to mention the always already present oppositional categories of man and woman)(See Du Bois 1998). Institutionally, a standardized method of obtaining information about individuals was not only necessary but pressing.
The Hollerith system worked so well for the United States in the 1890 census that other countries, notably Britain, adopted it for its tenth census of 1891 (NY Times, 1891). In fact, the New York Times coverage of the discussions for the adoption of the Hollerith system is quite favorable, at numerous points describing the previous census as containing many “blunders” and “shortcomings,” which, it implies, can be avoided with the adoption of the new American system. But more interestingly, the article goes on to explain why it is compulsory for Britain to have a robust system like the Americans: “But will the same blunders and shortcomings not be repeated in England in April[when the census was scheduled to take place]? That is the question which is troubling the average Briton when census matters are being discussed. America has been teaching England many useful things in the past twenty years, and it is thought probable that some at least of Mr. Porter’s crude ideas and slipshod methods may be imitated. After the thorough manner in which these have been exposed and ridiculed such a calamity is hardly to be expected, but the idea of London, like New York, waddling along for ten years under the stigma of a ridiculously incorrect statement of its population is enough to make the most stolid Englishman feel anxious and depressed (Ibid.).” As the article suggests, having an accurate census is important not just for accuracy but for nationalist pride and mental health. Therefore, to throw off the gathering and tabulation of accurate data is not simply undesirable but also unpatriotic. As the article says of the particular status of women in census data, they are possibly “the worst offenders against the requirements of the census, or at least against that particular requirement which requires them to tell their ages,” noting that no woman has yet to be put in prison for resisting to report her proper age (Ibid.).
The Times article also enlightens some of the key biopolitical aspects of the Hollerith system as it describes the stated intentions of the British Census Office for its use. One of them is the regulation of the poor, which some contemporary authors have suggested is precisely the function the public welfare system that used to exist in the United States (pre-Clinton), and which still exists across the European Union in divergent forms (See Piven and Cloward 1993). Prior to the census, statistics on the British poor were shoddily implemented by the Overseers of the Poor, whose positions were part of the Poor Laws of the early-17th century that provided food and clothing on a superficial level but were mostly means to discipline the lowest social strata. Specifically, the article states that one of the most important pieces of information is on the number family members living in a single apartment. Interestingly enough, it interprets this statistic as beneficial for the poor, as it will aid officials to go after real estate brokers and tenement owners who falsely advertise a space the size of a closet as a room, and tackle the “evils of overcrowding.” In light of this, along with governance reaching the level of life, we would also suggest that the Hollerith system contributes to the birth of the “social problem,” which a great deal of contemporary social science research is based on. One of sociology’s well-respected academic journals is aptly titled Social Problems .
One such instance of the production of a population as social problem was the Third Reich’s use of Hollerith’s system. As Gotz Aly and Karl Roth suggest, the statisticians that oversaw the project of tabulating populational statistics for the Third Reich were considered “soldiers of science,” that is, they were deemed to be just as valiant as those who were fighting on the front lines (Aly and Roth, 85-98). And indeed, as the British census had, the Nazi census of 1933 was perceived to be a response to security concerns and estimates for future food production, but was in fact the most modern and efficient means of social engineering and control, specifically targeting those of “non-German ethnic backgrounds,” which was nothing more than a coded label for Jews. As the adjoining image of an advertisement from 1933 shows, the Hollerith was seen as an all-seeing, panoptic eye, watching over the German peoples.
Hollerith on Hollerith
Therefore it is unsurprising that Hollerith, in a presentation to the Royal Statistical Society in England on 4 December 1894 regarding the success of this tabulation system, often refers to one distinguishing feature of his system from past systems of tabulating census data—data compilation (as opposed to collection). He writes, “[W]hile we classed our population as native white, foreign white, and coloured [in the 1880 census, on which he also worked], this was extremely unsatisfactory. . . Again it was apparent that if we wished to consider the progress of the negro[for example] in regard to illiteracy, we should know the number of illiterates at each age-period” (Hollerith, 678). In contemporary statistical language, Hollerith is suggesting that his system provided cross tabulations—a joint distribution of two or more variables, an all but impossibility with the older system used by the census. In this instance, he suggests that one could track literacy through the combination of race, age-period and literacy, which in contemporary social scientific research is called “intersectional” analysis. None of this could be possible without the quantification of the collected data. Additionally, the two operations of sorting and counting could be done simultaneously, or independently of the other (Hollerith, 680).
Moreover, the particular categories of identification (of which there was a total of twenty six) that Hollerith uses throughout his address to the RSS are as random as the usage of his system for the 1890 census. He consistently goes back to several “variables,” as it were, to make his point about the efficiency and superiority of his system of data tabulation. These are age; sex; birthplace of mother; “conjugal condition” [marital status], under which the option of “divorce” is so conspicuously left off; race; citizenship status; and occupation. One of Hollerith’s major selling points for his system is the degree of specificity that it could sustain thanks to its efficiency and sophistication. As he writes, “[f]or the foreign, in addition to the above, whether a citizen or alien, and whether the person could speak English. For the coloured, a distinction as to Black, Mulatto, Quadroon, Octoroon, Chinese, Japanese, and Indian” (Hollerith, 680). One does not have to venture far to see where these categories are remediated today. They can be found in the contemporary categories of the current census as well as the General Social Survey, on which a significant amount of quantitative social science training and research is done. Though this is most clearly what Foucault calls the statistical production of populations, how this process bears on identification and identity are significant as reflected in the raciology of American public discourse. For example, with “Black” and “Mulatto” as separate categories, Hollerith (and by extension the census at large) reproduces the fundamental tension of the American conception of race. It is at the same time reproducing the contemporaneous understanding of races as biologically separable, while also conceding to the possibility of race-mixing. For if one were to suggest that an individual could be both “black” and “white,” then it could be concluded that one should give up on the category entirely. But, in actuality, Hollerith’s example of the specificity achieved by his system(which he intended to be a sign of its superiority), quite simply demonstrates in its barest form, the way in which statistical calculation of populations produces the hegemonic epistemological infrastructure of society’s understanding of itself—the Foucaultian episteme.
Popular and Trade Press on Hollerith
If one were to be wary or even downright skeptical of Hollerith’s self-representation of the tabulation system, we could investigate the way in which popular weeklies and journals such as Harper’s and Scientific American portrayed the Hollerith system to see how it was portrayed not to a scientific community, as his address to Royal Statistical Society had been, but to a business community and also the reading public.
A pictorial advertisement in an August issue of Scientific American shows five images that are “slice of life” snapshots (figuratively, they are actually drawings) organized in continuous fashion. A mustached man—conceivably a “statistics clerk”—at a desk tabulating data; that is, punching in collected data into individual punch cards using the hand-puncher. A few pictures later, you see another image with a woman in the foreground seated at a desk among other people (men and women) also tabulating data. Next to her desk are stacks of paper, assumed to be already punched cards. Below that is an image of a group of men huddled around a table of punched cards that are being thrown at a pile of stacks on the floor, with a young man struggling to stack them properly. One imagines that this is supposed to signify the level of overwhelming efficiency of the electric tabulating system that overtakes even the ability for the young man to stack the books of punched cards.
The enthusiasm for the Hollerith system is spelled in greater detail in an article published in an issue of Harper’s Weekly published on 19 August 1899. It highlights the greater efficiency and accuracy of the system in conducting the census, boasting that it will allow the census to “embrace a greater area; for the first time the inhabitants of Alaska, Hawaii, and Puerto Rico are to be included in the count” (Harper's Weekly, 821). Additionally, it congratulates the decrease in labor that the Hollerith system provides, stating that “the improvement effected by the substitution of an automatic process in this work can be judged by the fact that one machine does the labor of twenty clerks under the old system” (Ibid.). And lastly, the article highlights the accuracy of the mechanical computation, which gives the census a level of trustworthiness that is not afforded to humans, who are capable of such things as error. For example, “if one of the details—say that of sex—is not punched, the electric plunger will not register, and the automatic bell at the side of the machine which announces the completion of the record will not ring. It is, then, a comparatively easy matter to go back and supply the missing information” (Ibid.). Lastly, it is worth noting a small detail in light of the suggestion in Foucault’s argument that biopolitics is characterized by the State’s intervention and interest in producing knowledge. The article reports that “the publisher will be the government; the publication will be designated as the Twelfth Census of the United States" (Ibid.).
The success of Hollerith’s electric tabulation system for the 1890 census paved the way for businesses to become interested in Hollerith’s Electric Tabulating Machine. One such industry was the burgeoning railroad companies. An article which appeared on 19 April 1895 in the Railroad Gazette, a trade journal, is indicative of the wave of interest in Hollerith’s system after the 1890: “A number of prominent railroad accounting officers have recently examined, with much interest, an invention for doing the great mass of the figuring in a freight auditor’s office by machinery, at a considerable saving in time and expense, and with perfect accuracy; and as the devices are exceedingly ingenious, and of interest to all accounting officers, whether they are likely to use them or not” (Railroad Gazette 1895). As is the case for many of the articles in trade journals, this article on Hollerith’s system reads almost like an advertisement, though with key points highlighting the way in which the system would be beneficial for railroad accounting officers in particular. What is useful to understand about the Hollerith system in this instance is the presentation to an industrial readership as a model of record-keeping and number crunching, which is in line with Hollerith’s own rendering of the benefits of his tabulation system in his address to the Royal Statistical Society. Moreover, record-keeping is one of the fundamental characteristics of Max Weber’s ideal-type of bureaucracy, which he argues is the dominant mode of social organization in modernity (See Weber 1914). Hence, the Hollerith system is the key turning point in not only the biopolitics of demographics but the rationalization of everyday life.
Herman Hollerith did not in fact file a patent for the punch cards that would bear his name, but rather for the machines used to inscribe, read and sort them. This section will describe the initial design of each of these technologies and their importance.
Hollerith punch cards were typically pieces of thin cardboard (although different forms of paper were utilized, with varying levels of success, throughout its history) cut to be 3.25 inches by 7.375 inches- the same size as the US currency in 1887, so that the cards could be stored in treasury bins. Hollerith cards initially had 24 columns, although later versions as many as 90. A corner of the punch cards was typically cut in order to ensure that they maintained the same orientation when stacked or tabulated, but held no further meaning.
These punch cards were, by admission of their inventor, conceptually based on several pre-existing technologies: the Jacqaurd Loom, the automatic telegraph, and train tickets punched by the conductor in holes used for description. The idea for the punch card tabulator was initially suggested by Dr. John S. Billings to Hollerith while both were working on the 1880 US census, stating that there “ought to be a machine for doing the purely mechanical work of tabulating population and similar statistics”, thinking that such a machine would create cards with the description of the individual notched into the side (Austrian). Hollerith’s first attempts at creating a tabulating machine involved punching holes into a roll of paper, with the relative position of the holes representing different aspects of the individuals recorded (i.e. sex, race, number of members of the household, etc). This role of paper could then be run over an electrified drum, which made electrical contact through the punched holes and activated the counters in a manner similar to the automatic telegraph. Yet the use of the roll of paper was ultimately unsatisfactory to Hollerith, for “the trouble was that if, for example, you wanted any statistics regarding Chinamen, you would have to run miles of paper to count a few Chinamen.” (Austrian). Drawing on his knowledge of the train tickets- so-called “punch photographs”- Hollerith abandoned the linear roll of paper for sets of punch cards- pieces of paper or cardboard cut into standardized shape, typically a rectangle- each of which could stand for an individual. Initially, Hollerith actually used a train conductor’s punch to create his cards, but the strain incurred from the repetitive exercise proved to be inhibitive.
The move from the strip of paper to the individualized punch card enabled the unit of information- for the census, the individual- to be “processed once, rearranged in new combinations, and processed again, until every bit of useful information was extracted” (Austrian). This shift away from the remediation of both the Jacquard Loom and automatic telegraph is of the utmost importance, for it not only eased the extraction of specific information (Hollerith’s Chinamen”), it fundamentally reconstituted the relationship between the data recorded and the statistician. While the rolls of paper could be read and counted, punch cards constituted a malleable assemblage that could be dismantled and reconstructed into all possible permutations. Remembering that the information that this invention was created to record was individual data- that each card represented an individual- one can see that the technology of the punch card allowed for an understanding of the US population as an infinitely divisible and rearrangeable mass. Although the input of information into the rolls of paper were not meant to be serial in order, the linear nature of the medium nonetheless created a form of order unintended and unwanted by Hollerith- to find a few Chinamen it was necessary to scan the entire population, for the Chinamen were literally inseparable from the cohesive whole. With the punch cards, it was possible to create relationships that did not necessarily exist amongst the individuals represented. All married female octoroons whose mothers were born outside of the country could be melded into a demographic group (quite literally by extracting and compiling their cards), regardless of whether or not these individuals were even aware of each other’s existence. They were, to reiterate Deleuze's term, "dividuals."
Cards were encoded by the placement of the holes punched into them. For this reason, the information written on the cards themselves (typically numbers, and often lines indicating column and row) were unnecessary to the technology, used only for quick correction or, for those well versed in the process, reading individual cards by sight (Austrian). To this purpose, a metal sheet cut the same size as the cards with the meaning of the individual holes could also be used to decipher the information held within(Austrian). Tellingly, during the 1890 census when Hollerith Punch Cards were first introduced, most of the cards were blank (Austrian). The data encoded into the punch cards at the dawn of its use can be thought of as being both digital and analogue. In a very basic sense, the presence or lack of holes is a binary system- the coded information is either “on” or “off”. Yet in its initial use, each position on the card stood for a single piece of information- one space was designated for male, another for married individuals, etc. As such, the coding was a process of spatializing analogue data. Actual binary cards were later produced (used primarily for computer programming) in which the holes represented zeroes and ones, and on these cards the relative position of the holes was of significantly less importance. Despite the increased sophistication of systems of encoding (the 80 column card with twelve punch positions, which came to be the dominant format known generally as the “IBM card”, used the eleventh and twelfth position as a “zone punch” which symbolized plus or negative signs, or other coded information which changed the interpretation of the other ten punch positions), most punch cards were used to store information in a spatialized analogue format. The transformation of information into a spatialized data enabled its decoding to be simultaneous- all information on a card could be read at once.
The material necessities of the punch card- that it be cheaply reproduced and easily punctured- resulted in an object that was, like other paper products, quite fragile. The phrase “do not fold, bend, spindle or mutilate” was a common phrase that accompanied punch cards that warned against abuse of the material which would render it illegible to the tabulating machines. This saying gained cultural currency in the 1960’s and 1970’s as representative of the alienation and dehumanization that accompanied the rise of computers, inspiring both a movie and a novel of the same title (needs citation). The phrase invoked the irony of the situation in which people were forbidden to violate material objects that were the very tools by which they were integrated into technocratic bureaucracies. Such prohibitions were not meant to be taken lightly, for the material disfigurement or destruction of the punch card meant potential disruptions of Kafkaesque proportions- taking the example of the 1890 census, since each individual was represented by a punch card, a card’s misuse or alteration fundamentally changed the way in which an individual counted, or if they counted at all. Once the use of punch cards spread throughout various institutional contexts, following the census in military and health care systems and then rapidly expanding through both the public and private sector, the physical integrity of the punch card often had a direct effect on the physical integrity of the individual.
The Hollerith punch card, along with similar technologies such as the Jacquard Loom, challenges the relationship of paper with the idea of "surface", what Jacques Derrida calls the "paper principle" (Derrida@48). Information is stored in the punch cards not through the inscription of traces on its surface, but rather in the alteration of its shape. Despite the admonishment not to "fold, spindle or mutilate" the punch card, the act of inscription is a permanent mutilation of the paper's surface. This puncture causes the two sides of the paper to meet, creating new edges and boundaries that are held internally, creating a lack that holds meaning. Unlike a surface that can be erased or wiped clean, the punch card becomes the information punched into it. The chad, the leftover piece of surface that has been removed by the puncture, is immediately discarded as refuse. Since meaning is held within absence, the presence of the chad is an unwelcome impediment that recalls residual possibility.
The "pantograph" or "Keyboard punch" was the artifact used to inscribe information into the Hollerith punch cards. The first iteration of the pantograph consisted of "a suitable frame carrying a card holder, a punch adapted to engage and punch a card held by said holder, mechanism for operating the punch, a keyboard supported on the frame, an index finger movable over the face of the keyboard" which controlled the punch (Patent #487,737). The keyboard on this machine was not what one would today typically imagine when hearing the term, but was rather a metal strip with holes that corresponded to the relative positions on the punch card. This keyboard could be replaced to create a new positioning system, and could also be removed and held up against any punch card to reveal its meaning. The meaning of "key" in this context should therefore be understood as an encoding and decoding mechanism rather than a button. During the 1890 census, specially trained clerks (the majority of whom were young women)would convert the data collected on handwritten forms into punch cards using this technology.
Of particular interest is the distance created by the pantograph between the operator and the punch card. Instead of directly punching holes into the card, the clerk instead pushes an index finger into holes in the metal keyboard which then causes a parallel action by the punching mechanism that then "engages" the card. On a purely practical level, it is possible to gain greater accuracy using the larger metal keyboard as opposed to punching the smaller punch card directly. Further, by punching the card close to the joint of the machine instead of at the edge grants the user greater leverage and thereby lesser strain from repeated use. More importantly, the "doubling" of action in the pantograph keeps the operator from directly interacting with the punch card- while the train conductor's punch was merely a tool, the pantograph was a nascent machine. Raw material (slips of paper) were inserted into the back of the machine, its "finger" was moved in pre-set motions, and the final product (punched card) was removed for further processing. Although the next iteration of the pantograph eliminated this divide, the estrangement of user and product presaged further developments of the technology in which the punching mechanisms were covered in a literal "black box", and only the user interface (which came to resemble modern keyboards) was visible.
Although often located in the same room, the technologies used to imprint punch cards and those used to read, sort and analyze them remained separate until the introduction of the IBM 604 in 1948 (needs citation). It is important to note that the sole purpose of the keyboard punch was to take data legible to humans and to transform it into data legible by a machine. Handwritten forms from which the data was transcribed could be counted and arranged in the same manner as the punch card- for each form also represented a singe person- but such a process was intensely consumptive of both time and resources (the 1880 census actually ran out of funds before finalizing its analysis of the population). Hollerith tabulating and sorting machines drastically cut the effort needed to conduct such an analysis, but in order to achieve this ends the from had to be transformed into the card. Since the pantograph is itself a machine of transcription that meditates and produces punched cards, it can be argued that the punch card is a medium of communication between machines. The pantograph receives input from the handwritten forms and internalizes this data in the form of the punch card, which is then read by the tabulator which then produces statistics. While this suggests that the keyboard punch is a form of rudimentary interface, because the two machines remained separate for the majority of the mediums' institutional dominance (punch cards were, for the most part, obsolesced in the 1970s) punch cards should be seen as an inherently machine-centric (as opposed to anthropocentric) medium.
An important turning point in the technology of the pantograph is the invention of the IBM Type 016 motor-driven Duplicating Key Punch in 1929. This machine enabled the duplication onto multiple cards the information held on one pre-punched "master" card. This development can arguably be seen as having effects on punch cards similar to those that print had on writing. The potential for what Elizabeth Eisenstein has dubbed "textual drift", i.e. inevitable errors or changes made in transcribing older copies, can be "fixed" through the production of multiple exact copies (Eisenstein, 88). This enabled the encoding and reproduction of complex computer programs on punch cards, for if a mistake was found the card could be duplicated up to the erroneous column, corrected by hand, then further duplicated. Although the invention of the duplicating key punch helped to assuage the problem of operator mistakes, incorrectly punched cards would plague the system from its inception through its current iterations. Several methods were developed to verify the proper coding of the cards, from Hollerith's system of two operators duplicating each other's work and a third party check to actual machines designed specifically for the purpose of correction. As was seen in the 2000 US presidential election, incorrectly punched holes and chads that refuse to accept their absencing continue to haunt the use of punch cards.
Tabulator and Sorter
The Hollerith tabulating and storage machines were two separate, yet connected machines that were ultimately integrated into a single device. The central piece of technology was a pin punch, which was a set of electrified pins that were pressed down onto the punch card. Where there was no hole, the pins would telescope back into the device, and where there were holes the pin would pass through the card into a small vial of mercury which would complete the electric current. Each pin corresponded to a counter held in the tabulator’s frame, which would advance by one unit every time an electric current was completed. At the same time, particular combinations of holes (determined prior to tabulating) would open different compartments on top of the sorter. In this manner, a clerk could simultaneously tabulate punch cards as well as sort them by different categories (such as by state, gender, or a mixture of the two) to be then rearranged and tabulated again. Developments in this technology led first to the integration of the two devices, then to the automation of sorting.
"Hear the Census with its bells
What a world of work
Their wild confusion tells,
How they klingle, klangle, klingle
In that Inter-Ocean room
Till your tympanums all tingle
At the jingle, jangle, jingle
And you wish you were at home
Keeping time, time, time,
To some other sort of rhyme
Than the tin, tin tabulation of those bells!"
|The Bells, Edgar Allen Poe|
Multiple Uses of the Punch Card Technology
The technology of Hollerith's Punch Cards were not limited to the collection of statistical data in the census. In fact, the technology was used for a multiplicity of purposes. In 1896 the Hollerith formed the Tabulating Machine Company. This was the first company dedicated to tabulating and accounting. (IBM Official Website) The company lent out the Punch Card machines to other companies, that needed to tabulate a large amount of data. The punch card equipment began to be used by the NY Central Railroad. The company was renamed International Business Machines (IBM) in 1924. Some significant technological transformations took place in this period. IBM created The Multiplying Punch Machine in 1931. This machine could read digits from a card, add, subtract, and multiply. (Patent 2178950) In 1937, IBM produced the first test scoring machine. The machine now has taken the form of the scan-tron that is still widely used in schools today. In 1954, the Electronic Data Translating System was invented. This was a system was an improvement upon the Punch Card system. It transformed intelligence on the datacards through photoelectric sensors. These sensors allowed for transference from punch cards to computing machines, quickly. In 1966, the Punch Cards were being used for Medicare identification, for the Social Securities Administration. In 1974 the Punch Cards were used for recording and Collecting data intended for betting and similar contests. (Patent 3906192)
Punch Cards used as Voting Machines
The Punch Card system has been used for tabulating votes because the equipment is portable, inexpensive, simple, and mechanical. The results gathered from the punch cards can be tallied quickly. In the 2000 election, 1/3 of all voters used the punch cards. The significance of the Punch Cards can be seen by the failure in the technology to adequately represent the votes in the 2000 election. Here voters left incompletely punched holes producing invalidated ballots. The argument was that people who were democrats and minorities, tended to live in districts that had the outdated punch-card system of voting, as opposed to republicans who had the newer more streamlined voting machines. (Who Uses Inferior Voting Technology?, by Stephen Knack; Martha Kropf PS: Political Science and Politics © 2002 American Political Science Association, p 541)
From Punch Cards to Magnetic Tape
In the 1940s there was a general trend in businesses to invest in machines that would save human labor. (The Impact of Automation on the Field of Accounting, by C. L. Keenoy The Accounting Review © 1958 American Accounting Association, p.2) The problem was solved through extending the punch card concept of integrated data processing where machines could be linked together, through a common language, an an automatic chain, automating different processes and operations. The integrated data processing concept resulted in the "sales-tronic" system. This system was first employed in retail, as there was a need for multi-variable flexible medium for recording information. There were factors that changed from transaction to transaction, such as mark-ups or discounts. This required a more flexible recording medium. (p.4) This was done with a punch paper tape recorder. All the information from the register is punched into paper tape. The data was then fed into electronic processing machines that had a tape reader. This notion of a strip of paper, was later improved upon by the magnetic tape recorder.
The most significant technological shift, was from the Punch Card, to the use of magnetic tape as a means of storing data. The Punch Card system anticipated the shift to the magnetic strip as a means of storing data. The development of cards as a means of storing data, was to ensure a more advanced system of collection and storage. It was more technologically advanced because it saved labor in the coding of information, transforming information into a pattern. The design of the Punch Cards was also labor saving in the retrieval of data. However the problem was that it was difficult to retrieve data, and every card was a unique instance of information. This produced massive stacks of cards, which was problematic for storage space. The system anticipated the storage of data by use of magnetic strips, because it eliminated the need for storage space. The transition to tape storage occured in 1952. The immediate reason for this technological change was because the Social Security System needed to solve their storage issues. The technology for the tape recorder was developed around the time that Punch Cards were invented. It was developed as a means of recording time, so that the hours of factory men, from the time and that they began work, to the time that they left, could be recorded. (Patent 400283) The magnetic tape was first adopted in the form of a reel. This was read linearly so that the access to the medium is sequential. This is in contradistinction to the Punch Card system, where a person could randomly retrieve data from any point. (Mullin, Jack, "Discovering Magnetic Tape," Broadcast Engineering, Intertec Publishing, Overland Park, KS, May 1979.) Thus it would seem that the Punch Card system was more advantageous in this respect. However, data-storage shifted from tape to disc. (IBM Official Website) With the disc it was possible to retrieve data from any point, thus invoking the logic of the Punch Cards.
Austrian, Geoffrey. 1982. Herman Hollerith, forgotten giant of information processing. New York: Columbia University Press.
Derrida, Jacques. 2005. Paper machine, Cultural memory in the present. Stanford, Calif.: Stanford University Press.
Eisenstein, Elizabeth L. 2005. The printing revolution in early modern Europe. 2nd ed. Cambridge ; New York: Cambridge University Press.