# The Encyclopedia Set

*The reference shelf as a first database: arranged, signed, bounded and already dated*

Kanwar Partap Singh Gill, MD · November 6, 2026 · Roots

The last article in the eleventh edition of the *Encyclopaedia Britannica* is about a word that had stopped working. It stands on page 1064 of the twenty-eighth volume, the one whose spine reads *Vetch to Zymotic Diseases*, and it runs to a single paragraph. *Zymotic*, it explains, comes from the Greek for ferment. The term had been used by Dr William Farr of the British Registrar-General's department for diseases that were "epidemic, endemic and contagious," on the theory that they arose from some morbid principle acting in the body in a manner analogous to fermentation. The entry lists the diseases the word came to cover: typhus and typhoid, smallpox, scarlet fever, measles, cholera, whooping cough, diphtheria. Then it delivers its verdict, in the same level voice it has used for twenty-eight volumes. The science of bacteriology had displaced the old fermentation theory, and the term "has practically dropped out of use."

I find it hard to think of a better last word for a great reference work. Some forty thousand headings, by the editors' own count, marched in alphabetical order toward a dead category of disease, and the encyclopedia, arriving there, said so. A reader who began at *A* and persisted to the end would close the final volume on a lesson the editors had not set out to teach: that knowledge is arranged by people, at a date, and that some of what they arrange is already on its way out of the world while the ink dries.

This essay is about the family reference shelf as a type, and about the encyclopedia set as the most ambitious object that could stand on one. It does not say which set, if any, stood in any particular house; its subject is the object and its history, and what that object taught about knowledge to anyone who learned to look things up before there was anywhere else to look. I call the set a first database on purpose. Long before the word meant software, a multivolume encyclopedia was a collection of records filed under keys, with an index, a plan of what counted as an entry, rules about who could write one, and a date stamped on the whole. Its users learned to query it. They also learned, if they were lucky, what it could not know.

## Spines

The spines are the first thing a set teaches, and they teach it before a single volume has been opened. The 1911 *Britannica* ran to twenty-nine volumes, the last of them an index, and each of the other twenty-eight announced its contents by its first and last headings, so that the row reads like a poem composed by the alphabet. *Medal to Mumps*. *Ode to Payment of Members*. *Payn to Polka*. *Shuválov to Subliminal Self*. *Submarine Mines to Tom-Tom*. The first volume runs from *A* to *Androphagi*, a people whom Herodotus placed among the man-eaters, and the second begins at *Andros*, a Greek island. Nothing connects the cannibals to the island except the order of their letters, and that is the point. Alphabetical order is a perfectly impartial filing rule, and like every impartial rule it produces strange company.

Anyone who has used an encyclopedia knows what that company does to attention. One goes looking for one thing, and the eye slides to the column beside it. The editors of 1911 called their method the dictionary plan, and they were frank about its double nature. A science studied in a treatise is a continuous argument, each chapter built on the one before. A science filed under the dictionary plan is broken into headings, and each heading is sent to its place in the alphabet, so that its history, its methods and its particular questions may end up volumes apart, each in the neighbourhood of strangers. From the scholar's point of view that is a scandal. From the browser's point of view it is a gift. A shelf arranged by the alphabet has no idea what you are looking for, and therefore cannot prevent you from finding something else.

The editors were not sentimental about the gift. In his long Editorial Introduction, signed in London on 10 December 1910, Hugh Chisholm explained why the eleventh edition had been written and printed all at once rather than volume by volume, as the ninth had been between 1875 and 1889. When an encyclopedia comes out over many years, he argued, the contents of each volume are "arbitrarily determined by the alphabetical order of headings," and by the time a later volume appears the world has moved, so that "the facts included in one volume may contradict those in another." He gave an example. It had seemed perfectly reasonable in 1879, he observed, to bring the article on the history of England no further than the accession of Queen Victoria, forty-two years earlier. "Up-to-date-ness," in his own inverted commas, was a demand the reading public had not yet learned to make.

The alphabet is not the only order, and it is not even a single order. Bhai Kahn Singh of Nabha spent about fourteen years, from 1912 until he finished the manuscript on 6 February 1926, compiling the *Gurushabad Ratnakar Mahan Kosh*, the great Punjabi encyclopedic dictionary of the words of the Guru Granth Sahib and of Sikh historical writing. Its 64,263 entries are arranged in the order of the Gurmukhi script, and the Gurmukhi alphabet, the *paintī* (ਪੈਂਤੀ, the thirty-five), does not begin as the Roman one does. It opens with the three vowel-bearers ੳ, ਅ and ੲ, then ਸ and ਹ, and only after them the ranks of consonants from ਕ onward, grouped by the place in the mouth where each is formed. A word beginning with *s* therefore lives near the front of a Punjabi lexicon, where an English reader expects to find the *a*'s and *b*'s. The neighbours are different because the script is different. A filing rule that looks like nature from inside one language looks like a convention from outside it, and a child who learns to look things up in two scripts learns, without being told, that order is something people make.

The *Mahan Kosh* has an Amritsar chapter of its own. Its printing began on 26 October 1927 at the Sudarshan Press in Amritsar, which belonged to the poet Dhani Ram Chatrik, and the four volumes of the first edition were finished on 13 April 1930. According to the account in Punjabi University's *Encyclopaedia of Sikhism*, its entries reach well beyond scripture: the histories and locations of gurdwaras, more than seven thousand words of Perso-Arabic origin printed in their own script, trees and herbs with their botanical Latin names, diseases and medicines, terms of music, prosody and philosophy. It is easy to admire it as a monument to one man's industry, and the same account marvels that he accomplished it single-handed. It is more instructive to notice what kind of database it is. Its key is the word as it occurs in a body of texts. Its records hold etymology, meanings by context, and quotations from the places where the word is used. It is built outward from a scripture and a literature. An English general encyclopedia is built outward from a notion of everything. The two shelves teach different first lessons about what knowledge is for: one says that knowledge is the understanding of particular texts that matter, the other that it is a survey of a world.

## Two Kinds of Authority

A set of matched volumes speaks in one voice. The typeface is uniform, the columns are uniform, the binding is uniform, and the effect of all that uniformity is to make every entry look equally settled. A paragraph resting on two centuries of measurement and a paragraph resting on one contributor's prejudice are indistinguishable on the page. This is the encyclopedia's quiet power, and its quiet danger.

The great nineteenth-century editors understood that a reader needs some way to tell such paragraphs apart, and their answer was the signature. In the ninth edition, as in the eleventh, many articles closed with the initials of their authors, and a reader could turn to the list of contributors and learn whose knowledge stood behind the prose. Chisholm quoted the ninth edition's editor, William Robertson Smith, on what those initials meant. They were not designed, Smith had written, "to lighten the responsibility of the editors." No editor could command every subject, but the staff had spared no effort "to secure the accuracy and sufficiency of every contribution." In the margin Chisholm named the principle: the two sources of authority. One was the name of the individual contributor. The other was the name of the *Encyclopaedia Britannica* itself, and his ambition in 1911 was to make the second more firmly established.

Two sources of authority sound reassuring until one reads what they jointly underwrote. The eleventh edition has an article headed SIKH. It is short, it is signed with initials, and it begins correctly: the word, it says, means "learner" or "disciple." It describes the distinction between initiated and uninitiated Sikhs and the outward signs of the initiated. Then its register changes, without any change in the type. "The Sikh is a fighting man," it announces, "and his best qualities are shown in the army, which is his natural profession." A sentence later comes a character sketch: "Hardy, brave and slow-witted, obedient to discipline, attached to his officers, he makes the finest soldier of the East." In peace, it adds, he "shows a decided fondness for money." Toward its close it counts the Sikhs in the Indian army, some thirty thousand, and remarks with approval that the sect is cherished by the military authorities, before listing the other imperial outposts where Sikh soldiers and policemen served.

None of this is offered as opinion. It sits on the page with exactly the calm of the paragraph on zymotic diseases, and the binding lends it exactly the same standing. Yet the sketch is plainly the view of an imperial army appraising its recruits, the doctrine that sorted the peoples of India into martial and unmartial kinds, set in the type of general knowledge. The entry on AMRITSAR, in the first volume, belongs to the same family. It gives the city's origin in the excavation of the sacred tank by Guru Ram Das, which is right, and then says that Ahmad Shah Durrani "routed the Sikhs at the great battle of Panipat" before he destroyed the temple. The battle of Panipat, in January 1761, was fought between Ahmad Shah and the Marathas; his great slaughter of the Sikhs, remembered as the Vadda Ghallughara, came in February 1762 on the road toward Barnala. The same entry assures its readers that the Golden Temple, taken as a whole, is "too squat to have much architectural merit apart from its ornamentation."

The children's encyclopedias of the same era were built on the same footing. Arthur Mee's *Children's Encyclopaedia*, issued in fortnightly parts between 1908 and 1910 and then sold for decades in bound sets, answered children's questions through a figure called the Wise Man in its "Book of Wonder." It also, by the account of later readers, took the British Empire's pride for granted, held Christianity to be the one true religion, and ranked the races of the world, however politely it asked that the lower ones be treated with respect. Wonder and hierarchy arrived in the same binding, and a child had no means of separating them.

Reading these pages now, as someone who grew up in the city they describe, I feel less anger than a kind of instruction. They show with unusual clarity how a reference work manufactures trust. The editors of 1911 were not careless; the edition was admired for its scholarship and earned much of that admiration. The trouble is that the form makes no visible distinction between the measured, the inherited and the convenient. A reader sent from *Sikh* to *Sikhism*, as the article directs, passes through identical typography whatever the quality of the knowledge beneath it. The conclusion to draw is narrower than distrust. Binding is a technology of trust, and that trust given to a form is different from trust earned by an argument. A uniform page is the most persuasive thing a publisher can make, which is exactly why a reader should ask what each paragraph on it is standing on.

## A Cross-Reference Followed Home

If the alphabet scatters knowledge, the cross-reference is the encyclopedia's attempt to gather it back. Ephraim Chambers, whose *Cyclopaedia* appeared in London in 1728, was the first compiler to use cross-references in large numbers and on a system, and his purpose, as historians of the form describe it, was to counter the fragmenting effect of alphabetical order: a reader sent from one article to another could reassemble a subject that the alphabet had dispersed. The French *Encyclopédie* of Diderot and d'Alembert began life as a project to translate Chambers, inherited his habit, and made it famous. Its *renvois* have been called ancestors of the hyperlink. They also acquired a reputation of their own, as the instrument by which the encyclopedists smuggled heresy past the censors.

The reputation comes partly from Diderot himself. In his article *Encyclopédie* he reflected on what cross-references could do, and among their possible uses he mentioned the satirical reference that quietly undermines an authority too powerful to attack openly, though he cautioned against using it often. Since at least 1759, readers have claimed that the work contained a whole system of such subversions, and one example turns up in almost every account. The article ANTHROPOPHAGES, on cannibals, observes that the pagans once accused the first Christians of eating human flesh, an accusation built on garbled notions of their mysteries, and then sends the reader onward: see EUCHARIST, COMMUNION, ALTAR. The joke seems obvious. Follow the arrows and the sacrament stands in the company of the man-eaters.

Follow the arrows further, though, and the joke dissolves. The scholar Marie Leca-Tsiomis pointed out that the same article, with the same cross-references, had already appeared in English in Chambers's *Cyclopaedia*, a work no one has suspected of anticlerical mischief. In the French it is signed with the letter G, the mark of the abbé Mallet, an orthodox clergyman who would presumably have objected had he seen an insult in his own translation. Her conclusion is that there is no evidence of any system of ironic cross-references in the *Encyclopédie* at all. The most celebrated example of encyclopedic subversion is a faithful rendering of an inoffensive English entry, and the story of its daring has been passed from one reference work to the next for more than two and a half centuries.

The genre has a gift for this kind of irony. The word *encyclopaedia* itself, by the standard dictionary account, comes from two Greek words, *enkyklios paideia*, a rounded or general education, which copyists of a 1470 Latin edition of Quintilian ran together as though they were one; the spurious single word passed into Latin and then into English, and readers ever since have imagined that the ancients had a genre they did not have. Chisholm's introduction contains a smaller specimen, buried in a footnote. When the eighth edition reprinted Sir Walter Scott's article on drama unchanged, in homage to its author, it also reprinted an older footnote referring readers to allusions "on p. 147," which were indeed on page 147 of the seventh edition and were on page 137 of the eighth. The cross-reference had been copied forward with the reverence due to the text and had quietly stopped pointing at anything.

I like these chains because they show both what a cross-reference is good for and what it cannot do alone. It takes the reader somewhere. It cuts a path through the volumes that the alphabet did not cut. But a path is also a story, and a story invites belief. The reader who follows *Anthropophages* to *Eucharist* feels the small thrill of discovery, and the thrill feels like evidence. What settles the question is a different motion altogether: not following the reference forward to the next entry, but tracing the entry backward to its source, its translator and its date. The eleventh *Britannica*, for reasons of its own, seems to have sensed the difference. It declined to clutter its pages with "mere cross-references" of the bare *see elsewhere* kind, and declared that the index volume, which carried them instead, "now acts in this respect as the real guide to the contents." The index was the query engine, and the editors wanted readers to use it rather than wander.

## Already Out of Date

Every encyclopedia begins to age while it is being printed, and the honest ones admit it. Chisholm's remark about the history of England halting at Victoria's accession is that admission in its Edwardian form. The century that followed tried several remedies. In 1902 and 1903, eleven supplementary volumes issued under the auspices of *The Times*, nine of them text and the others an atlas and an index, converted the ninth edition into the tenth without rewriting the main fabric. The eleventh was written and printed at once, so that at least its volumes would not contradict one another. In 1932 the publishers adopted what they called continuous revision, updating articles on a schedule and reprinting. Each remedy shortened the lag. None removed it, because a printed page cannot be revised after it has left the press, and a set on a shelf is a collection of pages that have all left the press.

For a physician the most instructive example of the lag sits in the third volume, under BERI-BERI. The article is careful and, by its own lights, fair. It describes the disease as a peripheral neuritis, maps its distribution from Japan to Brazil, and then lays out the rival explanations side by side. The cause, it reports, "is believed by many authorities to be an infective agent of a parasitic nature," though attempts to identify the agent have not been entirely successful. Another strongly supported hypothesis blames unwholesome diet, and the article gives the evidence: beri-beri was constantly prevalent among Japanese sailors until 1884, when their dietary was changed, after which the disease declined until it disappeared. Ronald Ross, it adds, suggested arsenical poisoning; a doctor in Sarawak blamed a fungus in mouldy rice. Every hypothesis is given the same typographical weight. The word that would have settled the matter does not appear, because it did not yet exist. When the Nobel committee honoured Christiaan Eijkman in 1929, its presenter recalled that the Dutch government had sent a commission to the Indies at a time when "bacteriology was in its hey-day," and had naturally looked for a germ; it was Eijkman's chance observation of laboratory hens falling ill on polished rice, and recovering when the bran was restored, that pointed elsewhere, and Casimir Funk who later gave such protective substances the name *vitamins*. The eleventh edition's last article celebrates bacteriology for displacing an old theory. Its article on beri-beri shows the same bacteriology, at the height of its prestige, steering many authorities away from the truth. Both stand in the same binding, written in the same voice, dated the same year.

The article was not careless. It was accurate to its date, and more scrupulous than many later summaries in admitting that the question was open. What it could not do was tell its reader which of its four hypotheses the next two decades would keep, and nothing on its spine or its page warned that the most prestigious of the four was the one about to fail.

There is a consolation in the lag, and it changes how an old set ought to be read. An encyclopedia that has fallen out of date does not become worthless. It changes genre. It stops being a reference and becomes a source. The eleventh *Britannica* is useless as a guide to the causes of disease and invaluable as a record of what educated Britain believed about disease, and about Sikhs, and about the architecture of Amritsar, in the years just before the First World War. Its article on zymotic diseases now documents the moment when bacteriology finished off an older theory; its article on the Sikh documents the confidence of the martial-race doctrine at its height. Read as testimony rather than as instruction, the same page that misleads becomes evidence, and the date on the title page, which the original reader could afford to ignore, becomes the most important thing about it. The difficulty is that a set on a shelf does not announce when it has crossed that line. The binding looks the same on the day it stops being a reference as on the day it was delivered.

Here the encyclopedia set has to be defended against its own admirers. It is now often invoked as the emblem of a whole way of growing up before the internet, as though every childhood had a row of matched volumes in a glass-fronted case and an adult nearby who said *look it up*. That picture is a class memory dressed up as a generational one. A multivolume English-language encyclopedia was expensive. In the twentieth century its American owners built the business on direct marketing and door-to-door salesmen, and what the salesman sold, beyond the volumes, was aspiration: the promise that this was what an educated household owned. In Punjab, as in most of the world, the shelf a household actually kept might hold a dictionary, a school atlas, a religious calendar, some textbooks handed down from an older child, or nothing at all, and the reference that answered most questions was a person: a teacher, a neighbour who read the newspaper, whoever in the lane had been to college. Language made a second barrier. A child whose first language was Punjabi met an English encyclopedia, if he met one at all, as a book in his second or third language, which is a different encounter from meeting it in the language of home.

I would rather keep the encyclopedia as a model than as a badge, and the model is more useful precisely because the object was never universal. A first database need not be bound in imitation leather. For many learners it was a library's reference room, consulted standing up. For others it was a volume like the *Mahan Kosh*, built to serve a literature rather than a general curiosity, or a well-thumbed dictionary that had to do the work of everything. The habit that matters, putting a question to an organized store of knowledge and learning how the store was organized, could be acquired at many shelves, and at some that were not shelves at all.

## A Different Promise

The *Britannica* gave up print in 2012. That March the company announced that the 2010 edition, thirty-two volumes and 32,640 pages, would be the last on paper. The announcement produced a small genre of elegies, most of them about the passing of a certain kind of authority, and most of them slightly beside the point. What ended was not authority. What ended was a particular promise.

A printed set promises a bounded whole. It says: here is what the editors judged worth knowing, in this many headings, on this plan, as of this year. The promise includes its own limits. A reader can see where the set stops; he can hold its edges in his hands and count them. The fifteenth edition of 1974 went further and printed its schema, a single volume called the *Propaedia* that laid out a hierarchical outline of knowledge, so that a reader could study the map before choosing a road. A search engine makes a different promise: abundance and recency, without edges. It does not say how much it holds, or on what plan, or as of when. Its order is relevance, which is a calculation about the person asking as much as about the material.

Neither promise is simply better. The bounded set was slow to change and, as the SIKH entry shows, could fix a prejudice in type for a generation. The unbounded index corrects some errors within hours and puts primary sources within reach of anyone with a connection, which no printed set ever did. What deserves to be carried from one to the other is not the binding but the habits the binding made visible. Who wrote this? Who checked it? When was it last revised? What kind of evidence stands behind it? An open online encyclopedia has, oddly, revived some of the nineteenth-century apparatus in a new form. Its revision histories and discussion pages expose the editing that a handsome set concealed, and anyone patient enough can watch an entry change its mind. The reader who learned to look for initials at the foot of an article has a head start in reading the history of a web page. Something has been lost as well, and it is worth naming exactly. A search box answers only the question it is asked. The spine of a volume, the column beside the one you wanted, the heading that happened to share your initial, offered the question nobody asked, and a good deal of a curious childhood was made of those. Links and recommendations offer their own accidents, but they are accidents arranged by a calculation about what the reader already likes, which is the opposite of the alphabet's indifference. The alphabet did not care whether *Androphagi* interested you. That indifference was a kind of freedom.

The word *database* repays a moment's attention here. A database is more than a pile of facts: it is a set of records shaped by a schema, retrieved by keys, and maintained under rules about what may be entered and how it may be changed. The old set taught all of this without naming any of it. Its headings were keys. Its index was a query interface. Its editorial policy was a schema with opinions. Its edition date was a timestamp, the most important field in every record and the one most readers ignored. Retrieval has since become immeasurably more powerful, and the questions a careful user should put to it have not changed in kind. What is the key? What is the schema? Who may write here? What is the timestamp? A child who once ran a finger down the spine labels to find *Medal to Mumps* was learning, under another name, to ask the first of those questions. The others take longer, and the set did not volunteer them.

## What a Reference Cannot Do

Flaubert spent his last years on a novel about two copy-clerks who try to learn everything from books, and he died before he could finish it. *Bouvard et Pécuchet* was published in 1881. Its heroes, having come into money, retire to the Normandy countryside and work their way through agriculture, chemistry, geology, history and much else, failing at each with tremendous diligence. Their medical chapter is the one a doctor reads through his fingers. They acquire a pasteboard anatomical model, take it to pieces, and cannot remember how to put the pieces back. They borrow volumes from the local physician, who lends them with the assurance that they will never reach the end; they take notes on the cases of longevity, obesity and extraordinary constipation in the *Dictionary of Medical Sciences*, and regret that they have never encountered such marvels as the famous gluttons, the ossified man and the old mayor of Angoulême whose nose weighed three pounds. They read that one authority estimated the force of the heart as sufficient to lift a hundred and eighty thousand pounds and another at about eight ounces, and they conclude that physiology is the romance of medicine. In the English translation of 1904 the narrator adds the sentence that might stand as the novel's epigraph: "As they were unable to understand it, they did not believe in it."

Then a pedlar sells Bouvard the *Health Manual* of François Raspail, whose doctrine has the charm of simplicity. All diseases proceed from worms, and the remedy is camphor. The two friends take camphor as snuff, chew it, hand it round in cigarettes, and begin to treat their neighbours. At a farmer's bedside they meet the physician, Vaucorbeil, and the argument that follows is one of the sharpest pages in the book. Pécuchet cites authorities; the doctor tells him to let systems alone; Pécuchet asks whether he is an empiric, then; the doctor answers that he goes by observation; Pécuchet asks what happens if one observes badly; and the doctor replies that to start with it is necessary to have practised. When Vaucorbeil finally appeals to his diploma, Pécuchet answers that a diploma is not always an argument, and the doctor storms off threatening prosecution for the illegal practice of medicine. Flaubert lets neither side win. The clerks are ridiculous, but the physician's credential is not much of an answer either. The translator adds a footnote with a sad fact in it: Raspail had been called in to attend Flaubert's sister Caroline before her death in 1846. Flaubert was a surgeon's son. He knew what was at stake when a household turned from a doctor to a book, and what was at stake when a doctor answered a question with a certificate.

What the novel dramatizes, with Flaubert's merciless patience, is a distinction that no reference shelf can draw for its reader. Take beri-beri again, through its levels of knowledge. A general encyclopedia gives an overview: what the disease looks like, where it occurs, which explanations were in favour when the entry was written. A specialist literature gives argument: the case for the germ and the case against it, the reasons one hypothesis prevailed and the others were abandoned. Primary evidence gives the observations on which the argument rests: the navy's changed rations and its vanishing cases, hens that sickened on polished rice and recovered on bran, the experiments that followed. And practical judgment, which no level of text supplies, decides what to do for the particular person in the room, whose numbness and swollen ankles are not an entry. Bouvard and Pécuchet leap from the first level to the last without passing through the middle two, and that is their real folly, worse than any belief in worms. The overview gave them a vocabulary. They mistook it for a warrant.

The encyclopedia's honest function is orientation. It tells a beginner where a subject lies, roughly what its parts are called, and which way to walk next. That is a great deal, and anyone who has stood bewildered at the edge of a new field knows how much. It is not evidence, and it is not judgment, and its uniform voice makes the difference easy to forget. Learning what a reference cannot do is part of learning to use one, and it is the part most often skipped, because the reference itself never mentions it. The *Britannica* of 1911 did not print, at the foot of its article on beri-beri or on anything else, a line saying *this is an introduction; the argument is elsewhere; the decision is yours*. Vaucorbeil's grumble, that one must first have practised, is the missing line, and it is not one that a book can say on its own behalf.

## The Shelf and Its Successor

George Eliot gave one of her doctors his vocation from a row of dusty volumes of an old cyclopaedia on the highest shelf of a small home library, and [another essay on this site](../the-day-i-first-thought-i-might-become-a-doctor/index.html) reads that scene more closely. The novel that follows is a long test of what such a beginning is worth. A reference shelf can set a mind moving. It cannot tell the mind where to stop believing it.

What remains, when the set has gone to the second-hand dealer, or never existed in the first place, is a habit of seeking orientation and a sense of where orientation ends. I want to keep both, and I want to keep a third thing the old shelf taught almost by accident: that every reference system makes editorial choices about what deserves an entry, and that the choices are visible only to someone who goes looking for them. The *Mahan Kosh* decided that the vocabulary of a scripture and its literature deserved sixty-four thousand entries and fourteen years of one scholar's life. The eleventh *Britannica* decided that forty thousand headings would hold the knowledge of its world, and found room among them for an army's character sketch of a people and a connoisseur's sniff at their holiest building. Every database since has made decisions of the same kind, about which records to create, which fields to keep, whose word to trust and when to stop updating, and very few of them print those decisions on the spine.

The set's last article remains, for me, the best lesson in reading all the others. *Zymotic diseases*: a heading kept in the alphabet after the theory that produced it had died, presumably because readers still met the word in old registers and reports and needed to know what it had meant. The editors could have quietly dropped it. Instead they filed it in its place, explained it, and told their reader in plain words that it no longer described the world. That is the most a reference can do with its own obsolescence, and it is more than most references do. A first database that ends by marking one of its own records as out of date has delivered its most important lesson on its final page: that knowledge kept in order is still knowledge kept by someone, at some time, and that the time belongs to the fact.

## Evidence and Editorial Note

**Documented findings.** The essay relies on the author's public biography: upbringing in Amritsar; Punjabi as first language, with Hindi, Urdu and English; work as a physician. The 1911 *Encyclopaedia Britannica*: the list of twenty-nine volumes and their spine ranges (*A–Androphagi*, *Andros–Austria*, *Medal–Mumps*, *Ode–Payment of Members*, *Payn–Polka*, *Shuválov–Subliminal Self*, *Submarine Mines–Tom-Tom*, *Vetch–Zymotic Diseases*); the article "Zymotic Diseases" (vol. 28, p. 1064) with Farr's usage and the statement that the term had practically dropped out of use; Hugh Chisholm's Editorial Introduction (dated London, 10 December 1910), including the ninth edition's publication over many years, the defects of volume-by-volume publication, the 1879 history of England, "up-to-date-ness," some 40,000 headings against some 17,000, the two sources of authority and Robertson Smith's statement on initials, the 1902 supplementary volumes, the index as "the real guide to the contents," and the footnote on Scott's article and the p. 147/p. 137 reference; the articles "Sikh" (vol. 25) and "Amritsar" (vol. 1), quoted as printed: the Wikisource transcription of the 1911 edition. The battle of Panipat (14 January 1761, Durrani Empire against the Marathas) and the Vadda Ghallughara (5–6 February 1762, on the route from Kup-Rahira toward Barnala): Wikipedia articles "Third Battle of Panipat" and "Vadda Ghallughara." The *Mahan Kosh*: Dharam Singh's entry "Gurushabad Ratanākar Mahāṅ Kosh" in *The Encyclopaedia of Sikhism* (Punjabi University, Patiala; online text) and the Wikipedia article "Mahan Kosh" (64,263 entries in Gurmukhi alphabetical order; work from 1912 to 6 February 1926; printing at Sudarshan Press, Amritsar, owned by Dhani Ram Chatrik, 26 October 1927 to 13 April 1930; scope of entries). The order of the Gurmukhi alphabet: standard reference knowledge. Arthur Mee's *Children's Encyclopaedia* (fortnightly parts 1908–1910, the "Book of Wonder," its imperial and religious assumptions): Wikipedia article "The Children's Encyclopædia." Chambers's *Cyclopaedia* and systematic cross-references, the *Encyclopédie* as an expanded translation, Diderot's remarks on satirical cross-references, the claim of a "system" since 1759, the article "Anthropophages" and its cross-references, and Marie Leca-Tsiomis's finding that the article and its references were already in Chambers and were signed by Mallet: Wikipedia article "Encyclopédie," section on cross-references, citing Leca-Tsiomis. The etymology of *encyclopaedia* from *enkyklios paideia* and the copyists of a 1470 edition of Quintilian: Wikipedia article "Encyclopedia," citing the *American Heritage Dictionary*. *Britannica* publishing history (direct marketing and door-to-door sales in the American era from 1901, continuous revision from 1932, the 1974 *Propaedia*, the 2012 end of print, the 2010 edition of 32 volumes and 32,640 pages): Wikipedia article "Encyclopædia Britannica." The 1911 article "Beri-Beri" (vol. 3), including the infective hypothesis held by "many authorities," the Japanese navy's change of diet in 1884, Ross's arsenical hypothesis and the mouldy-rice theory: Wikisource transcription. Eijkman, the Dutch commission, the remark that bacteriology was in its heyday, the hens fed polished rice and cured by bran, Grijns, and Funk's naming of vitamins: Nobel Prize in Physiology or Medicine 1929, award ceremony speech and Eijkman facts page (nobelprize.org). Flaubert's *Bouvard and Pécuchet*: the 1904 English translation (M. Walter Dunne; Project Gutenberg), including the Auzoux manikin, the *Dictionary of Medical Sciences*, the heart estimates, "As they were unable to understand it, they did not believe in it," Raspail's *Health Manual* and camphor, the exchange with Vaucorbeil, and the translator's note on Raspail and Flaubert's sister; publication in 1881 and Flaubert's father's profession are standard reference knowledge. **Vendor representations.** None.

**Recollection and tradition.** The author's upbringing in Amritsar is part of his published biography. The essay does not say which encyclopedia, if any, the author's household owned or consulted, and it describes no remembered act of reading, browsing or looking up. The reference shelf is treated as a type. Remarks about what households in Punjab might keep on a shelf are general observations, not claims about any family. Sikh history is cited from the sources above; nothing is asserted about what all Sikhs or Punjabis believe.

**Analysis and forecast.** The description of the encyclopedia as a first database, the readings of the 1911 articles, the contrast between bounded and unbounded promises, the account of levels of knowledge, and the interpretation of Flaubert's novel are the author's interpretation as of November 6, 2026. **Independence.** This essay is not sponsored by any institution, company, health system or technology developer. It is personal reflection, not medical advice; readers seeking care should contact their own physician's office.



## Related reading

- [Where Knowledge Lived: Amritsar, 1995](https://personal.kpsgill.com/essays/where-knowledge-lived-amritsar-1995/)
- [The Teacher Who Made Chemistry Make Sense](https://personal.kpsgill.com/essays/the-teacher-who-made-chemistry-make-sense/)
- [The Library Desk](https://personal.kpsgill.com/essays/the-library-desk/)