Still in the Book

Nostalgia was invented as a disease in 1688, and everyone knows it stopped being one. Forty-nine reference books, asked the same question in their own words, show the first half of that happening in print. The second half is not true. Here is this year's code book.

<mainTerm><title>Nose-picking</title><code>F98.8</code></mainTerm>
<mainTerm><title>Nosomania</title><code>F45.21</code></mainTerm>
<mainTerm><title>Nosophobia</title><code>F45.22</code></mainTerm>
<mainTerm><title>Nostalgia</title><code>F43.20</code></mainTerm>
<mainTerm><title>Notch of iris</title><code>Q13.2</code></mainTerm>

The story of nostalgia is one of the few pieces of medical history that has escaped into general circulation. A Swiss medical student named Johannes Hofer, defending a dissertation at Basel on 22 June 1688, needed a Greek name for the thing his countrymen called Heimweh, and made one out of nostos, the return home, and algos, pain. For most of two centuries after that it was a disease in the ordinary sense: it had symptoms, a course, a prognosis, a place in the classification, and a body count. Then, the story goes, medicine stopped believing in it, and the word drifted out of the clinic and into the record shops.

The first half is true, and it is more interesting than the summary, because you can watch it happen sentence by sentence in the books themselves. The second half is false, and the way it is false is the reason this page exists. Nostalgia did not leave. It was demoted, one rubric at a time, over a hundred and twenty-seven years, until it became a thing you look up rather than a thing you have.

The shelf

Forty-nine volumes: thirty-six general medical lexicons, one specialist dictionary of psychological medicine, eleven official nomenclatures, and one army's medical history, printed between 1787 and 1981. Each was downloaded, its scanned text pinned by hash, and asked one question: is there an entry for nostalgia, and what does it say?

Pick a year. The entry you get is not typed out here; it is cut straight from that book's own scan, damage included. Where the run-in from the previous entry is showing, it is greyed.

Read along the shelf and the shape is not subtle. In the four oldest lexicons here, Quincy's Lexicon Physico-Medicum among them, the word does not appear at all; a century after Hofer coined it, the standard English medical dictionary had no use for it. It arrives in the first decade of the nineteenth century already fully furnished, with a cause, a course, and two species. It swells. In the Dunglison editions of the middle of the century the entry runs to more than a hundred words, carries as many as eight other headwords redirecting into it, and ends on a clause no modern dictionary would print: that the wasting it produces is sometimes attended by hectic, which may speedily induce death. Then, in the space of about fifteen years, it collapses to a single word.

Four books could not agree where to file it

A disease has to go somewhere. Before there was one official list there were rival ones, and Hooper's Lexicon-Medicum of 1822 does something accidentally generous: it reprints several of them, one after another, as an appendix. Nostalgia turns up in four of those tables, in the same volume, under four different numbers.

This is worth sitting with before the rest. The disagreement people imagine is between an old credulous medicine and a modern sceptical one. It was never that tidy. In 1822 nostalgia was simultaneously a local disease of disordered appetite, a species of madness, and a hallucination, depending on which page of one book you had open.

The army that counted it

The strongest evidence that this was a real diagnosis, in the sense that it changed what people wrote down, is that an army counted it. The United States Surgeon General's Medical and Surgical History of the War of the Rebellion tabulates every reported case of every disease in the Union armies, and nostalgia is a printed line in the summary tables, in Class IV, Local Diseases, Order I, Diseases of Nervous System, between inflammation of the spinal cord and neuralgia.

Five thousand five hundred and forty-seven cases, and seventy-four deaths. Those two totals are worth stating carefully, because they are the source of a small confusion in circulation. The volumes report the figures in two separate tables, one for white troops and one for the United States Colored Troops, and they never add them up. Secondary accounts quote sometimes 5,213 and 58, sometimes 5,547 and 74, as though these were competing estimates. They are the same data: 5,213 plus 334, and 58 plus 16. Nobody is disagreeing with anybody.

And then the document does something that any honest account of it has to include, because it undercuts the use the figures are usually put to. The 1888 volume's own chapter on nostalgia opens by playing the disease down. Its wording is that nostalgia cannot be said to have been of frequent occurrence, as only 5,213 cases were reported among the white troops, or 2.34 cases annually per thousand of strength, and it goes on to note that epilepsy has more claim than nostalgia to be called a disease of our war camps, for 9,029 cases were reported. The book that certifies nostalgia as a camp disease of the war spends its first paragraph explaining that there was not very much of it. If you have seen the claim that nostalgia killed more men than enemy fire, the document usually cited for it is the one that refutes it.

Struck out where nobody struck it out

Here is where the received story goes wrong, and where it goes wrong is instructive.

The Royal College of Physicians of London maintained an official list of what a disease was and what it should be called. Its title page describes it as subject to decennial revision: a standing committee re-deciding, every ten years, the contents of medicine. Six editions of it are on this shelf, from the first in 1869 to 1931, along with the United States Marine Hospital Service's list of 1878. Nostalgia appears in none of them. Not once, in seven official nomenclatures across sixty-two years, including the 1884 revision published four years before an American army was still tabulating the disease by name.

It is tempting to call that the moment nostalgia was struck off. It is not, for a plain reason: it was never in. The first edition, 1869, does not have it either. There is no decision to date, because the British nomenclature simply never admitted the word, while British and American physicians went on using it in their dictionaries, and one of those dictionaries, Tuke's Dictionary of Psychological Medicine of 1892, gave it a long signed clinical article that states flatly that nostalgia must always be defined as disease. In 1892 the word was, at the same time, absent from the official British nomenclature, alive in the leading British psychiatric dictionary, and a line item in the American army's statistics.

Every one of those absences is reported here with a control, because an absence read off a scan is only worth as much as the scan. Select any dashed volume on the shelf above and it will show you how many times that same file contains melancholia, insanity, neuralgia, hysteria, fever and delirium. If the medical vocabulary is legible and the word is at zero, the zero belongs to the book.

The ladder

Meanwhile, on the other side of the Atlantic and then in Geneva, nostalgia was being admitted to the classification that eventually swallowed all the others, and it has never left it. Each rung below is a document fetched, hashed and searched; the phrase quoted under each is a slice of that document.

A hundred and twenty-seven years, seven revisions, one direction of travel. It starts as an included term under mental alienation, beside lycanthropy. It gets its own standalone code in the American standard nomenclature. It is moved into the psychoses. The World Health Organization inherits it and files it among the situational reactions, next door to the code carrying combat fatigue. By the current revision it is an index term redirecting to adjustment disorder, unspecified.

Nothing here is a struck-through line. There is no year, no committee minute, no obituary. What happened to nostalgia as a diagnosis is what happens to a word in a bureaucracy that never throws anything away: it kept its entry and lost its meaning, one cross-reference at a time, and its neighbour through all of it, from the World Health Organization's first index in 1949 to the file the CDC published last June, has been a notch of the iris.

What the word was doing while this happened

The clinical sense was being displaced, in ordinary printed English, by the sense everyone now means. That much is not a discovery: corpus linguists have studied nostalgia's shift, and the historians of the concept have described it for sixty years. What is offered here is a small dated measurement of it, with its instrument fully exposed.

Twenty-eight phrases were fixed in advance, before any of them were fetched, in four groups. Phrases only a disease produces: cases of nostalgia, died of nostalgia, symptoms of nostalgia. Phrases only a feeling produces: a wave of nostalgia, a twinge of nostalgia, tinged with nostalgia. And the two objects of the longing kept apart: a place you are away from, against a time you are after. Each was measured in Google Books, decade by decade.

The decade is solid and the year is not, and it is worth being explicit about which is which. Read on raw yearly figures the sentimental register passes the medical one for good in 1928; smooth over five years and it is 1922; over nine, 1921. Drop any single one of the fifteen phrases and the answer moves by at most a year, so no one phrase is carrying it. What that adds up to is a decade, the 1920s, and an honest refusal to name a year inside it. The same decade, on the same test, is when longing for a time overtakes longing for a place.

The caveats travel with the number. Google Books is not a sample of English, and its composition changes over time, so a rising line can be a rising shelf rather than a rising usage. A phrase is a proxy for a sense and cannot tell you that a case of nostalgia in 1950 is a joke and in 1870 was not. One of the twenty-eight probes, attacked with nostalgia, returned nothing at all and is reported here rather than quietly dropped, which is the only reason to fix the list before fetching it.

A second, independent instrument agrees about the direction and adds something the first cannot. Counting pages in the Library of Congress's digitised American newspapers, and dividing by the number of pages digitised from that decade so the corpus growth cannot fake a trend, the word goes from 86 pages per million in the 1810s to 35,101 per million in the 1960s.

What is not mine, and what I could not check

The history of nostalgia as a medical concept is a developed field and almost nothing narrative on this page is new. Jean Starobinski's essay of 1966 established the modern account; George Rosen revived the subject in 1975; Michael Roth wrote the nineteenth-century French medicine; Helmut Illbruck wrote the intellectual history, including the chapter on the Swiss cowherd's melody said to have been banned on pain of death, a story which traces to Rousseau's Dictionnaire de musique of 1768 and for which no military order has ever been produced; Susan Matt wrote the American arc; Thomas Dodman's What Nostalgia Was is, in its publisher's own words, the history of clinical nostalgia from the coining of the term to its removal from medical discourse; Krystine Batcho settled the 1678-versus-1688 dating confusion; Fuentenebro de Diego and Valiente Ots framed it as a transient illness that was relayed into other diagnoses rather than deleted; David Anderson and Frances Clarke wrote the Union Army. Charlotte Taylor and Anna Marchi have taken the semantic shift to Hansard and The Times, which are better corpora for the question than the one used here.

What appears to be new is narrow, and it is the apparatus rather than the argument: the edition-by-edition sweep with its absences controlled, the mechanical attribution of the four filings in Hooper 1822, the reconciliation of the two Civil War figure-sets as one dataset, and the classification ladder assembled end to end and checked against the current release rather than against the memory of it.

Things this page could not settle, listed because the list is part of the result:

• ICD-11. The World Health Organization's current international revision requires an authenticated API and could not be searched here. Every rung above stops at the American clinical modification, which is a real code set in real use, but it is not the same document.
• The 1938 fifth revision of the international list. No fetchable full copy was found. Nostalgia is in the fourth and in the sixth, so it is very probably in the fifth, and this page does not say so.
• The 1920 code number. One scan of the American standard nomenclature reads it 14121 and another reads it 11131. The leading digits of that table are not legible in either scan, and no page image was read.
• Two scans on the shelf are too damaged to cut an entry from, though the word is present in them; they are marked as such rather than skipped.
• The mid-century gap. Every medical dictionary located between 1944 and 1980 on the Internet Archive is lending-restricted, so the shelf jumps from 1943 to 1981.
• Hofer's dissertation. Only the title page was read here, and only for the date, the coinage and the names on it. The Latin argument itself was not read; the Basel and Munich digitisations are both open and that is a separate evening. The standard English translation, Anspach 1934, is paywalled at JSTOR and undigitised at the Wellcome, so it was not consulted either.
• The newspaper series is thinner than intended. loc.gov answered 503 to almost everything, so the six control words that were supposed to sit beside nostalgia in that chart were never fetched. The script is committed with them still in it.