The Problem With Math Quote Collections
I spent about three years building a reference database of mathematics quotes by famous mathematicians while working on a pedagogical project. Most online lists you find are wrong, misattributed, or completely stripped of context. I discovered this the hard way when I was verifying quotes for a curriculum and realized half of what was circulating online had been invented or tangled together by people who found the wording attractive without checking the source. A properly curated collection requires tracking down original publications, correspondence, or verified secondary sources. The process is tedious and unglamorous. You will spend more time debating whether a particular quote actually came from Gauss than you will spend typing up the final list. Most people don't realize this before they start.
Mathematics Quotes By Famous Mathematicians
Here are some of the more reliable ones, with context attached because context matters more than you think. Pierre de Fermat wrote annotations in the margins of his personal copy of Diophantus' Arithmetica. The most famous one, concerning the impossibility of solving x^n + y^n = z^n for n greater than 2, includes the phrase "I have discovered a truly marvelous proof of this, which this margin is too narrow to contain." We have no evidence he ever actually found such a proof. The quote is real. The footnote is part of what makes it useful for understanding how Fermat thought about problems, not as a motivational poster. Paul Erdős had a system for ranking mathematicians based on creative output. He called it the Book, a mythical volume where God keeps the most elegant proofs. Erdős would say "which book?" when someone presented a particularly clean argument. This is a well-documented cultural artifact among number theorists. It tells you more about Erdős's personality and his standards than it does about any specific theorem.
Henri Poincaré wrote about intuition in mathematical creation. In his book Science and Method, he described how mathematical discovery involves selection among countless combinations, guided by aesthetic feeling. The full passage is longer and more nuanced than anything you will see quoted from it. Reading the original gives you something substantially different from the soundbite versions floating around. Kurt Gödel's incompleteness theorems changed how mathematicians understood their own field. His personal notes and correspondence reveal a man who was deeply concerned with the foundations of mathematics and skeptical of certain philosophical positions, including logical positivism. Quotes from his letters and private discussions carry more weight than those from popularized summaries because they show the actual reasoning behind his published results. John von Neumann was known for brutal directness. One commonly cited remark attributed to him about topology being "the group theory of the twentieth century" is likely apocryphal. I spent several months tracking down sources for this and couldn't find it in any of his published works, lecture notes, or verified correspondence. He said many things, but this particular phrasing doesn't appear to be one of them. Be careful about what you attribute to von Neumann.
Get the Full Details

Maryam Mirzakhani, the first woman to win the Fields Medal, gave an interview where she discussed the relationship between mathematics and creativity. She spoke about the process of working on a problem as similar to walking through a dark room, slowly finding your way. This kind of quote captures something genuine about mathematical practice that formal publications never convey.
How to Build a Reliable Collection
The main difficulty is verification. A quote appearing on a quotation website doesn't make it true. I learned to check against original publications first, then reputable secondary sources like academic biographies, collected papers, and documented interviews. Wikipedia entries are useful starting points but unreliable as final sources. I encountered a specific problem with a quote attributed to Andrew Wiles about proving Fermat's Last Theorem. The version circulating online said something inspirational about perseverance. The actual statement from Wiles was far more measured and technical, discussing the specific obstacles in the modular forms approach. When someone asked me why I had included the inspirational version, I had to remove it. The misattribution was widespread and persistent. I still encounter it. Another edge case involved a quote from Srinivasa Ramanujan about infinite series. Multiple versions existed, and I could not determine which was the original. I ended up excluding it rather than include an uncertain attribution. This is a decision you will face repeatedly. Being conservative about what you include is better than being comprehensive and wrong.
The practical workflow is straightforward but time-consuming. You collect candidate quotes, verify each against primary sources where possible, note the context in which each was said, and flag any that remain uncertain. A proper collection should include citations or at least references to where each quote can be found. I built mine in a spreadsheet with columns for the quote text, attributed author, source, confidence level, and notes on context. The spreadsheet became the working document, and the final publication was extracted from it.

Common Mistakes to Avoid
The biggest mistake is treating these quotes as decoration. They are not inspirational wallpaper for office walls. They are historical documents that reveal how mathematicians understood their own work. A quote from Emmy Noether about structural thinking means something different when you understand the actual mathematical revolution she participated in than when you read it isolated from that context. Another mistake is over-reliance on anecdotal collections. People remember the striking phrases and forget the qualifications. I found that some of the most useful quotes in my collection were the quiet ones, the offhand remarks that mathematicians made in seminar discussions. These often contained more genuine insight than the polished statements prepared for public consumption. You will also run into the problem of translation. Many mathematicians wrote in German, French, or other languages. A quote translated poorly can mean something entirely different from the original. I worked with several collaborators who could read primary sources in their original languages to verify translations. This made a significant difference in accuracy.
Where to Find Verified Sources
The most reliable starting points are collected works and correspondence volumes published by academic presses. Books like The Correspondence of David Hilbert, or the collected papers of various mathematicians, provide authenticated texts. Biographies by historians of mathematics, particularly those who specialize in primary source research, are also useful. Journal articles in History of Mathematics or Isis sometimes publish newly discovered or verified quotations. For living or recently deceased mathematicians, recorded interviews, oral history projects, and lecture recordings are valuable. The American Mathematical Society and the London Mathematical Society have oral history collections. These are not easily searchable but worth consulting if you are doing serious work. There is no single comprehensive database of verified mathematics quotes. That is a gap in the available resources. What exists is fragmented across biographies, collected papers, and various curated lists of varying reliability. Building a trustworthy collection requires assembling these fragments yourself and being willing to exclude material that cannot be verified.
The effort is worthwhile because the quotes themselves are genuine artifacts of mathematical culture. They show the people behind the theorems, the doubts, the humor, and the specific ways of thinking that formal mathematics suppresses. A well-documented collection of Mathematics Quotes By Famous Mathematicians can serve as both a reference and a historical document, provided it meets the standard of verification I described. Anything less is just noise.
