The Lost Notebook: Treasures Rediscovered
The remarkable story of Ramanujan's final mathematical discoveries
Learn about Ramanujan's Lost Notebook discovered in 1976, containing formulas he wrote on his deathbed that continue to yield new mathematical insights.
The Lost Notebook: Treasures Rediscovered
In the spring of 1976, the American mathematician George Andrews was visiting Trinity College, Cambridge, hunting through the papers of G. N. Watson for material on Ramanujan. Watson had spent much of his career working through Ramanujan's published notebooks. After his death in 1965, Watson's mathematical estate had been deposited at Trinity's Wren Library. Andrews asked to see the Watson papers, sat down in the reading room, and began to open boxes.
In one of them he found 87 unbound sheets of paper covered in a handwriting he recognized immediately. The sheets had never been published. They had apparently never been catalogued. They had been sitting in Watson's study for the better part of forty years, and after Watson's death they had travelled quietly to the Wren Library in an unmarked box. The handwriting was Ramanujan's. The formulas were dated to his last year of life, 1919 and 1920, in Kumbakonam, when he was already dying of what doctors now believe was hepatic amoebiasis. Andrews spent the afternoon reading. Within an hour he understood what he was holding. This was the manuscript the world had been hearing about for decades and had given up looking for. It was, as he would later call it, Ramanujan's Lost Notebook.
The discovery was immediately compared in the mathematical press to the finding of a lost Beethoven symphony. The comparison was not hyperbole. The manuscript contained roughly 600 formulas, most of them previously unknown, and many of them in Ramanujan's characteristic form of a bare statement with no proof. They belonged to the last year of a life that had produced, in roughly a decade of active work, a body of results that mathematics is still catching up to. The Lost Notebook was the final layer, and it turned out to contain the deepest results Ramanujan ever wrote down.
What the Lost Notebook Actually Contains
The Lost Notebook is not a bound book. It is a stack of loose sheets, roughly 135 pages in the Narosa facsimile edition published in 1988, densely packed with identities, continued fractions, partial theta functions, and a category of objects that Ramanujan called, in his last letter to Hardy in January 1920, mock theta functions. The mock theta functions are the centre of the Lost Notebook and the reason its discovery mattered so much.
In the 1920 letter, written from his deathbed, Ramanujan described these objects in a single suggestive paragraph. He listed seventeen of them. He gave some of their series expansions. He stated, without proof, that they had certain subtle transformation properties that were like, but not identical to, those of the classical theta functions Carl Gustav Jacobi had studied a century earlier. And then, before he could explain what he meant, Ramanujan died, at age 32, on 26 April 1920, in a house in Chetput outside Madras.
For the next eighty-two years, the mock theta functions were the most famous unsolved mystery in Ramanujan's work. Hardy thought they were real but could not systematize them. G. N. Watson worked on them for years and made substantial progress without being able to say what they fundamentally were. The Lost Notebook, when it was found in 1976, contained dozens more of them, with hints at identities connecting them, but still no definition, still no proof, still no framework. The objects were clearly coherent. They clearly belonged to some larger theory. Nobody knew what the theory was.
The Manuscript's Long Silence
The provenance of the Lost Notebook is itself a story about how easily important work can be misplaced. After Ramanujan's death in April 1920, his wife Janaki gave his final papers to the University of Madras. The university sent them to G. H. Hardy in Cambridge as part of the memorial effort to collect all of Ramanujan's work. Hardy looked at them, eventually passed them to his Cambridge colleague G. N. Watson in the 1930s, and thereafter the sheets seem to have become part of Watson's personal mathematical papers. Watson worked on them for years. When he died in 1965, his widow donated his papers to Trinity College, Cambridge. R. A. Rankin, a mathematician who examined the collection before it was sent to the Wren Library, later recalled seeing the Ramanujan sheets and identifying them as important. They were boxed up with the rest of Watson's estate and deposited at Trinity, where they waited for Andrews.
Eleven years separate the death of Watson and Andrews's discovery. The manuscript was not actually lost in the strict sense. It was sitting in a library that every serious number theorist in Britain knew how to find. It simply had not occurred to anyone that the loose sheets in Watson's papers were the final manuscript of Ramanujan. That is how the most consequential mathematical document of the twentieth century can spend eleven years in plain sight, not because it is hidden, but because nobody looks.
From Discovery to Understanding
The discovery of the Lost Notebook in 1976 was only the first act. The second act was the multi-decade project of actually understanding what was on the pages. Bruce Berndt at the University of Illinois took up the work in the late 1970s and, over the next four decades, produced a series of five volumes titled Ramanujan's Lost Notebook in collaboration with Andrews, each one supplying proofs and context for hundreds of the formulas. The last volume in the series was published in 2018. Berndt's editorial project is one of the longest sustained scholarly engagements with a single mathematician's papers in modern history, and it is still not finished.
The third act was the eventual understanding of the mock theta functions themselves. In 2002, a Dutch doctoral student named Sander Zwegers, working at the University of Utrecht, finally put the mock theta functions inside a proper framework. His thesis showed that each mock theta function is the holomorphic part of an object called a harmonic Maass form, a category of function that had not been defined in Ramanujan's lifetime and was only fully formalized in the late twentieth century. Zwegers had solved, 82 years after Ramanujan first wrote the word, the mystery that had stood at the centre of the Lost Notebook.
The fourth act is still unfolding. Kenneth Ono, working with Zwegers and others after 2002, has shown that the mock theta functions appear in string theory, in the formulas that describe the entropy of certain black holes. Results Ramanujan wrote on his deathbed, in a Tamil village, now describe the mathematical structure of objects at the frontier of twenty-first century physics. Ono has called this connection the spookiest experience of his mathematical career. It is hard to disagree.
What the Lost Notebook Means
The Lost Notebook is the document that proved Ramanujan was not simply a prodigy who happened to work in a neglected corner of early twentieth-century mathematics. He was, in the strict sense, seeing things that the rest of the field had not yet invented the language for. When the language finally arrived, eighty years later, his formulas were waiting for it.
The deeper lesson is that mathematical truth does not have a timetable. A young clerk in Madras, dying of disease, writing on loose sheets in a house he would not leave, could see into a territory that the collective mathematical community would not enter for four generations. The sheets waited in a library. Scholars eventually picked them up. New language eventually caught up to them. New physics eventually reached for them. What mattered was that the truth was on the page. Everything else was patience.
Back in the Wren reading room at Trinity, Andrews closed the box. The handwriting he had recognized within seconds had been waiting in unmarked sheets for forty years. The shelf had been patient. So had the page.
Key figures
Srinivasa Ramanujan
22 December 1887 to 26 April 1920, Erode and Kumbakonam (Tamil Nadu), then Cambridge, then Kumbakonam
George E. Andrews
Born 1938, Salem, Oregon, USA. Active from 1960s onwards at Penn State University
Bruce C. Berndt
Born 1939, Saint Joseph, Michigan, USA. Active from 1970s onwards at the University of Illinois Urbana-Champaign
Case studies
George Andrews Finds the Lost Notebook (Spring 1976)
In the spring of 1976, George Andrews, a 37-year-old American mathematician already known for his work on partitions and q-series, arrived at the Wren Library of Trinity College, Cambridge. He had come specifically to look through the mathematical estate of G. N. Watson, who had died eleven years earlier. Watson had been one of the most devoted interpreters of Ramanujan's notebooks, and Andrews was hoping to find unpublished manuscripts or correspondence that might illuminate some of the formulas he had been working with. The librarian brought out several boxes from the Watson bequest. Andrews sat down in the reading room and began to open them, one at a time. In one of the boxes, under other papers, he found 87 unbound sheets of loose paper. The first thing that struck him was the handwriting. It was not Watson's. It was instantly, unmistakably, the neat small Tamil-inflected English script of Srinivasa Ramanujan. The formulas on the sheets were dated to Ramanujan's last year of life. They were not in any of the published notebooks. They were, it appeared, an entire previously unknown manuscript, sitting in a library, unread for the better part of five decades.
Andrews spent the afternoon in the reading room. By the time he left, he had understood what the sheets contained. He recognized results Ramanujan had mentioned in his final 1920 letter to Hardy, including the mock theta functions that had eluded every mathematician who had tried to systematize them. He recognized partition identities, continued fractions, and modular-form-like objects, many of them previously unknown to the mathematical literature. Within days he had written to colleagues and begun arranging for the manuscript to be photographed and studied. The news spread quickly. The comparison to finding a lost Beethoven symphony was not journalist embellishment. It was the reaction of working mathematicians who understood that a body of Ramanujan's final work, written at the height of his insight and the depth of his illness, had just been handed back to them after half a century.
The rediscovery of the Lost Notebook reshaped an entire field. Within a decade, partition theory, q-series, and modular forms all had new problems to work on, drawn directly from the manuscript. The Narosa facsimile edition of 1988 made the sheets themselves available to any reader. The Andrews and Berndt critical volumes, beginning in 2005 and continuing through 2018, supplied proofs for the formulas. New doctoral theses, new research programmes, new conferences, and ultimately new physics all traced their origins to the afternoon Andrews spent in the reading room. And the most haunting detail is the smallest one. The sheets had not been lost in any real sense. They had been in the Wren Library the whole time, catalogued under Watson's papers, available to any scholar who asked to see them. The difference between present and found was not distance. It was attention.
Important work does not always announce itself. Sometimes it sits in an unmarked box in a library that hundreds of qualified readers walk past. The question is not whether you have access. The question is whether you arrive with the right set of questions and the right training to recognize what you are looking at. Andrews found the Lost Notebook not because he was lucky but because he had spent a decade thinking about the problems it solved. When he opened the box, he was ready. If there is unfinished work in your own field that you suspect is lying around in plain sight, the only way to find it is to do enough preparation that you will recognize it when it appears.
The Lost Notebook consists of 87 unbound sheets, roughly 135 pages in the 1988 Narosa facsimile edition, and contains approximately 600 formulas. It spent roughly 56 years unread between Ramanujan's death in April 1920 and Andrews's discovery in spring 1976, including at least 11 years of that time sitting in the Wren Library of Trinity College, Cambridge.
Bruce Berndt's Forty-Year Editing Marathon
In the late 1970s, shortly after George Andrews's discovery of the Lost Notebook, a young number theorist at the University of Illinois named Bruce Berndt made a decision that would define the rest of his career. He would work through Ramanujan's notebooks, formula by formula, and supply a proof for every one of them. Berndt was 38 years old at the time. He had already established himself as a specialist in modular forms and analytic number theory. What he proposed was to set aside most of his independent research and devote himself, indefinitely, to the most patient kind of mathematical housekeeping: reading Ramanujan's bare statements, figuring out what they meant, and constructing the derivations that Ramanujan himself had not bothered to write down. Nobody at the time thought of this as a career-defining choice. Berndt began with the earlier Notebooks, then extended the project to the Lost Notebook, and then kept extending it, one chapter at a time, for the next forty years.
The task Berndt took on is not the kind of work that wins prizes quickly. It is not the discovery of a new theorem, but the slow, unglamorous work of giving existing theorems their proper justification. In Indian traditional terms, what Berndt was doing is best described as upapatti, the supplying of the demonstration that the Kerala school astronomer-mathematicians insisted must accompany every result. Ramanujan had stated without proving. Berndt would prove, for decades, without stating anything new. The discipline this required is hard to overstate. A single Ramanujan formula could take weeks of work to place in its proper context, to connect to known theory, and to derive. Berndt worked through thousands of them. The five volumes of Ramanujan's Notebooks (covering the Notebooks) were followed by five volumes of Ramanujan's Lost Notebook (covering the 1976 discovery). Ten volumes, four decades, one sustained act of commentary.
By the time the fifth and final volume of the Lost Notebook series was published in 2018, Berndt had produced what is widely recognized as the definitive modern commentary on Ramanujan's work. Every formula now has a published derivation. Every chapter has been placed in the context of the mathematics that grew up after Ramanujan. The volumes are used as research references at every major number theory department in the world. Doctoral students routinely cite Berndt's proofs as the starting point for new research. The remarkable thing is that Berndt's work has not exhausted the Lost Notebook. New research continues to be published on formulas he had already treated, because each formula turns out to connect to further questions. Berndt provided the foundation. The field is still building on it.
Some of the most valuable scholarly work is the kind that looks, from the outside, like mere housekeeping. Someone has to supply the proofs that genius leaves unwritten. Someone has to thread together the bare formulas into a coherent exposition. Someone has to be willing to spend four decades on a single patient project. The modern academic incentive structure rarely rewards this kind of work in the short term, but in the long term it is often what allows a discipline to actually use a body of results. If you find yourself drawn to meticulous, long-horizon editing or commentary work, do not apologize for it. Bruce Berndt's career is the demonstration that such work can be as original and as consequential as the original discoveries it serves.
Bruce Berndt's ten-volume editorial project on Ramanujan's notebooks (five volumes on the Notebooks published between 1985 and 1998, and five volumes on the Lost Notebook published between 2005 and 2018) spans 33 years of sustained publication and roughly 40 years of total work. Each volume is several hundred pages of dense mathematical commentary and is the result of collaborations with a rotating cast of co-authors including George Andrews, Ken Ono, and many of Berndt's own students.
Sander Zwegers Cracks the Mock Theta Functions (2002)
In 2002, a Dutch doctoral student named Sander Zwegers, working under Don Zagier at the University of Utrecht, submitted a thesis titled simply 'Mock Theta Functions'. The thesis was short by the standards of mathematical dissertations, barely a hundred pages. It was also one of the most consequential PhD theses in the history of the field. Zwegers showed that every mock theta function Ramanujan had listed in his 1920 letter to Hardy, and every additional mock theta function he had recorded in the Lost Notebook, could be understood as the holomorphic part of a newly-defined class of objects called harmonic Maass forms. In a single piece of sustained argument, Zwegers had supplied the conceptual framework that Ramanujan had written his formulas inside of without ever being able to describe. Eighty-two years after Ramanujan's death, the central mystery of the Lost Notebook was finally solved.
The remarkable thing about Zwegers's thesis is not that it found a framework. It is that the framework was available. Harmonic Maass forms had been slowly developing in the modular forms literature since the 1950s, but nobody had connected them to Ramanujan's mock theta functions. It took a young mathematician working in the right tradition, with enough distance from the historical baggage around Ramanujan, to simply try the connection. When he did, it worked. Every mock theta function turned out to fit. Every transformation property Ramanujan had hinted at turned out to be derivable. The framework was not imported with force. It snapped into place. This is the shape a genuine solution takes in mathematics. When an old puzzle finally yields, the yielding feels less like a conquest and more like a door opening from inside.
Zwegers's thesis became instantly canonical. Within two years, Ken Ono, Kathrin Bringmann, Amanda Folsom, and a generation of younger number theorists had begun building on it. An entire new subfield, the theory of mock modular forms and harmonic Maass forms, grew up around it. New identities were derived. Old Ramanujan formulas that had resisted explanation suddenly made sense. And the Lost Notebook, which had been a precious but partly opaque artifact, became a living research text. A twenty-first century graduate student studying mock modular forms is studying, at one remove, the final mathematical insights of a dying man in a Tamil village in 1919 and 1920, now finally readable in their own proper language.
Some problems do not get solved until the tools to solve them have been quietly built in a neighbouring discipline. Ramanujan wrote the mock theta functions in 1919. The theory of harmonic Maass forms took another seventy years to mature. Zwegers put the two together in 2002. If you are working on a hard problem and making no progress, it is worth asking whether the language you need is being built, right now, somewhere else in mathematics or in an adjacent field. Sometimes the right move is not to push harder on the problem. It is to stay alert to the neighbouring work that might eventually become the key. When the key arrives, the door you had been pushing against for eighty years will open easily.
Sander Zwegers, 'Mock Theta Functions', PhD thesis, Universiteit Utrecht, 2002, supervised by Don Zagier. The thesis is approximately 100 pages long. It is now recognized as one of the landmark doctoral theses in modern number theory and is the founding document of the theory of mock modular forms.
Ken Ono and the Black Hole Connection (2011 onward)
In 2011, the American number theorist Ken Ono, working with his colleagues Kathrin Bringmann, Amanda Folsom, and others, noticed something strange. The mock theta functions that Sander Zwegers had finally explained in 2002 kept appearing in physics papers on black hole entropy. Specifically, the formulas that string theorists used to count the microstates of certain classes of charged black holes in type IIB string theory turned out, after sufficient simplification, to be the same formulas Ramanujan had written down in his Lost Notebook in 1919. Ono checked the identifications carefully. The connection was not a coincidence of surface appearance. The mock modular forms that Ramanujan had discovered were, up to specific technical equivalences, the generating functions for the microscopic degrees of freedom of certain black holes.
The philosophical force of this connection is hard to overstate. Ramanujan lived in Kumbakonam, Cambridge, and again in Kumbakonam. He had no exposure to general relativity, which was published in Einstein's full form only in November 1915, the year after Ramanujan arrived at Cambridge and six years before his death. He had no exposure to quantum mechanics, which was still a decade in the future. He had certainly no exposure to black holes, which would not enter theoretical physics as a serious object of study until after his death, and would not be recognized as physical entities with measurable entropy until the 1970s. And yet the formulas he wrote down in 1919, on loose sheets in a house in Tamil Nadu, turn out to describe with precision the counting problem at the heart of black hole quantum mechanics. In the Indian view that Ramanujan himself held, this is not strange. The mathematical structure of reality is one, and the devotee who can see it is seeing the same thing whether she is counting partitions of an integer or microstates of a black hole. The West has a harder time saying this out loud, but the data keep pushing in that direction.
The mock-theta-to-black-hole connection has been developed in a long series of papers from 2011 onwards, with contributions from Ono, Bringmann, Folsom, Atish Dabholkar, Sameer Murthy, and many others. It has become a standard topic at the intersection of number theory and string theory. The physical significance of Ramanujan's formulas is no longer speculative. It is concrete, citable, and embedded in active research programmes. Ken Ono has called the discovery of these connections the spookiest experience of his mathematical career, and in public lectures has described opening Ramanujan's Lost Notebook and finding, in the handwriting of a dying man in 1919, what now appears to be the counting formula for the entropy of a black hole. The Lost Notebook, in other words, is not a historical document. It is a piece of frontier physics research.
The deepest mathematical results often have applications that could not possibly have been anticipated by their discoverers. Ramanujan wrote down the mock theta functions because he saw them, and then he died. A century later, they describe black holes. This is not evidence of prophecy. It is evidence of something more interesting, that the mathematical structure of reality is unified enough that a deep insight into one corner of it tends, in the long run, to illuminate many other corners at once. When you do fundamental work, you are adding to a common store whose uses you cannot predict. This should not worry you. It should encourage you. The full value of your best work is rarely visible at the moment you do it, and that is not a defect. It is the nature of the material.
Representative papers include K. Bringmann and K. Ono, 'The f(q) mock theta function conjecture and partition ranks', Inventiones Mathematicae 165 (2006), and later work by Atish Dabholkar, Sameer Murthy, and Don Zagier on 'Quantum Black Holes, Wall Crossing, and Mock Modular Forms' (2012 and subsequent papers). The connection between Ramanujan's mock theta functions and black hole state counting is now a recognized subfield at the intersection of number theory and string theory.
Historical context
From Ramanujan's Death to the Twenty-First Century (1920 to the present)
Ramanujan's final year was spent in Kumbakonam in the Tanjore district of what is now Tamil Nadu. He had returned from Cambridge in March 1919 in very poor health. The house in Kumbakonam where he wrote the Lost Notebook still stands. It is a modest two-storey structure on Sarangapani Sannidhi Street, across from the great Sarangapani temple. The family was rooted in the orthodox Vaishnava tradition of the region, and Ramanujan himself remained throughout his life a devotee of Namagiri Thāyār, the family goddess at Namakkal, whom he credited with the appearance of his formulas. The modern context of the lesson is that the same house, the same temple, and the same devotional tradition are all still alive and can be visited. The distance between the Lost Notebook on the palm-like paper of 1919 and the current global research programme on mock modular forms is shorter, in space and in culture, than the distance between any comparable historical mathematical document and its current use.
The Lost Notebook is the document that makes clear that Ramanujan was not merely an unusually talented mathematician of his generation. He was a mathematician operating with a different relationship to his source material than most of his contemporaries. The formulas he wrote in his last year have turned out, repeatedly, to anticipate structures that the rest of mathematics would not invent for eighty or a hundred years. The question of how this is possible is still open. The data that make it a real question, rather than a piece of mythology, are the pages of the Lost Notebook.
Living traditions
The Lost Notebook is now a living research text rather than a historical curiosity. Active research programmes at Emory University (Ken Ono's group), the University of Illinois (Bruce Berndt and collaborators), Penn State (George Andrews), Utrecht, and many other universities continue to work through its formulas. The Ramanujan Journal, founded in 1997, publishes new results arising from the notebooks. SASTRA University's annual Srinivasa Ramanujan Prize honours mathematicians under 32 who work in areas influenced by Ramanujan's legacy. Internationally, conferences on mock modular forms routinely reference specific page numbers of the Lost Notebook in their programmes. The manuscript that sat unread for 56 years has, in the half-century since its rediscovery, become one of the most actively cited sources in contemporary number theory. That trajectory, from loose sheets in a Kumbakonam bedroom to the frontier of string theory research, is itself the most honest tribute to the man who wrote it.
- Ramanujan's House, Sarangapani Sannidhi Street, Kumbakonam: The modest two-storey house where Ramanujan grew up, where he wrote many of his earliest notebooks, and where he returned to die in 1919 and 1920. The house has been acquired and preserved as a memorial by SASTRA University and is open to visitors. Inside, the rooms have been restored to approximately their appearance in Ramanujan's lifetime. The small upstairs room where he worked on the Lost Notebook in his final months is particularly affecting. Across the street stands the great Sarangapani temple, which was part of the devotional fabric of his family life.
- Namagiri Thāyār Temple, Namakkal: The temple of Namagiri Thāyār, the family goddess of Ramanujan's lineage. The Namakkal Narasimha temple complex is carved into a hillside and houses the shrine of Namagiri that Ramanujan credited as the source of his mathematical visions. For any student of Ramanujan, visiting Namakkal is the closest one can come to the devotional context in which the Lost Notebook was written. The walk up the hill to the shrine is a small pilgrimage in its own right.
- Wren Library, Trinity College, Cambridge: The library where George Andrews rediscovered the Lost Notebook in the spring of 1976. The Wren Library is open to visitors for limited hours during term time and preserves the Watson collection among its mathematical holdings. For a serious visitor, the reading room itself is the historical site. This is the physical space in which a dying man's final mathematical thoughts, recorded in Tamil Nadu in 1919, were finally read by a living mathematician in 1976.
Reflection
- In your own life, have you ever created or written something that had to wait for the right audience or the right framework before it could be understood? What did the waiting feel like, and what finally unlocked it?
- The Lost Notebook sat unread in a Cambridge library for at least eleven years. What does it mean about mathematical knowledge, and about truth in general, that it can remain perfectly intact through long periods of complete inattention?
- Ramanujan wrote results in the Lost Notebook without proofs, attributing them to Namagiri Thāyār. Eighty years later, other mathematicians supplied the proofs and showed that every formula was correct. Who should be said to have discovered these results: the man who saw them, or the people who eventually understood them?