Abakcus — Book Lists
Warm-ups, Victorian classics, Martin Gardner, Raymond Smullyan's logic and a last shelf for people who want to suffer. 23 puzzle books in the order I would open them.

Tap a cover to open the book.
23
books, in the order to read them
5
steps, from warm-ups to puzzles that fight back
137
years between the oldest puzzles (1885) and the newest (2022)
7
with Martin Gardner's name on the cover
A puzzle is the only kind of mathematics that people do for no reason. Nobody is examined on it and nobody is paid for it. You see a question that looks as if it ought to be easy, it turns out not to be, and half an hour later you are still sitting there with a pencil. I think that half hour is the purest form of what mathematicians mean when they say they enjoy their work, and it is available to anyone who can count.
It is also older and more serious than it looks. Probability began with a gambler's puzzle about an unfinished game. Graph theory began with a puzzle about seven bridges in Königsberg. The questions in this list about weighing coins, crossing rivers and colouring maps have a way of turning into whole subjects a century later, which is why so many working mathematicians say they started with a book by Martin Gardner.
There are thousands of puzzle books and most of them are the same hundred puzzles reshuffled. So I kept to the books the others borrow from, and arranged the 23 of them in five steps. The first step is for warming up. The second goes back to the Victorians who invented the modern puzzle. The third belongs to Gardner alone, the fourth to Raymond Smullyan and logic, and the fifth is for readers who have finished the rest and want something that fights back.
One piece of advice before you start. Do not read these books, work them. Cover the answer, give the puzzle a real try, and give up later than you want to. A puzzle you solved after two days is worth more than fifty you looked up. If you would like something to do with your hands while you think, I have instructions for folding a hexaflexagon and for cutting a tangram.
In a hurry
“I have never opened a puzzle book”The Moscow Puzzles No. 01
“I want the classics everyone copies”536 Puzzles and Curious Problems No. 08
“I want one Martin Gardner book”The Colossal Book of Mathematics No. 15
“I like logic more than arithmetic”What Is the Name of This Book? No. 16
“A gift for a clever twelve-year-old”Math Games with Bad Drawings No. 04
“I want puzzles that hurt”Winkler's Mathematical Puzzles No. 20Before the books
Puzzle 1 · A warm-up
A snail sits at the bottom of a well ten metres deep. Every day it climbs three metres, and every night it slides back two. On which day does it reach the top?
The eighth day, not the tenth. After seven days and seven nights the snail is seven metres up. On the eighth day it climbs three more, reaches the rim, and there is no night to slide back.
Puzzle 2 · A classic
Each letter stands for a different digit, and no number begins with zero. Find the digits that make the sum SEND + MORE = MONEY correct. Henry Dudeney published it in the Strand Magazine in 1924.
9567 + 1085 = 10652. Start with M, which must be 1 because it is the carry from adding two four-digit numbers. Then S has to be 9 and O has to be 0, and the rest follows one letter at a time.
Puzzle 3 · An aha
Cut two diagonally opposite corner squares off a chessboard, leaving 62 squares. You have 31 dominoes, each of which covers exactly two neighbouring squares. Can you cover the board?
No. Every domino covers one dark square and one light square, so 31 dominoes cover 31 of each. The two corners you removed were the same colour, which leaves 32 squares of one colour and 30 of the other.
Puzzle 4 · A logic puzzle
On an island every inhabitant is either a knight, who always tells the truth, or a knave, who always lies. You meet two of them, A and B. A says: “At least one of us is a knave.” What are A and B?
A is a knight and B is a knave. If A were a knave the statement would be true, and knaves cannot say true things. So A is a knight, the statement is true, and the knave it speaks of has to be B.
Step I · Books 01–05
Five friendly books that ask for nothing beyond school arithmetic. Any one of them will tell you within an evening whether you are a puzzle person.
No. 01 · Step I
Boris A. Kordemsky, edited by Martin Gardner · 1954, English edition 1972
Boris Kordemsky was a mathematics teacher in a Moscow high school, and the puzzle book he published in the 1950s became the most popular one ever published in the Soviet Union, reprinted again and again. Martin Gardner edited the English translation in 1972, adapting a few puzzles for American readers and leaving the Russian flavour of the rest alone.
What makes it the right first book is the slope. It opens with riddles a ten-year-old can do and climbs so gradually, through matchstick puzzles, dominoes, cross-number problems and magic squares, that you hardly notice you have started doing number theory near the end. The answers at the back are full explanations, not just numbers. It costs about as much as a sandwich.
Best forComplete beginners of any age. This is my first recommendation.
No. 02 · Step I
Alex Bellos · 2016
Alex Bellos sets a puzzle every other week in the Guardian, and he is also the author of Here's Looking at Euclid. Here he has chosen 125 puzzles from the last two thousand years and done what few puzzle books bother to do, which is tell you where each one came from. You learn who first asked how to get a wolf, a goat and a cabbage across a river, and how a question from a Singapore school exam about a girl called Cheryl and her birthday went around the world in a weekend in 2015.
The puzzles are grouped by type, logic, geometry, practical problems, word games and numbers, and each chapter runs from easy to quite hard. Bellos writes the solutions as conversations, showing the wrong turns as well as the right one. It is the best modern anthology I know, and a good second book after Kordemsky.
Best forReaders who want the history of each puzzle along with the puzzle.
No. 03 · Step I
Ian Stewart · 2008
When Ian Stewart was fourteen he started a notebook for interesting bits of mathematics that he was not being taught at school. He kept adding to it for the next fifty years, by which time he was a professor at Warwick and one of the best-known popular writers on the subject, and this book is the notebook emptied onto the table.
It has no chapters and no order. A logic puzzle sits next to a short account of a famous theorem, then a number trick, then a joke about mathematicians. The puzzles have answers at the back and the rest is there to be read. You can open it anywhere, which makes it the ideal book for a bedside table or a long flight. If you finish it there is a second volume, the Hoard of Mathematical Treasures.
Best forBrowsers, and anyone who wants puzzles mixed with stories and jokes.
No. 04 · Step I
Ben Orlin · 2022
Strictly speaking this is a book of games and not of puzzles, and I include it because a good game is a puzzle that changes every time you play it. Ben Orlin, the teacher behind the blog Math with Bad Drawings, collects seventy-five and a quarter of them, by his own count, and nearly all need only paper, pencils and a second person.
You will know Dots and Boxes. You may not know Sprouts, which John Conway helped invent, or Ultimate Tic-Tac-Toe, where every move decides which small board your opponent must play in next. For each game Orlin gives the rules, a sample round drawn in his proudly terrible stick figures, and then a few pages on where the mathematics is hiding. He writes the funniest prose in the genre, and I have also listed his calculus book.
Best forFamilies, teachers and anyone who would rather play against a person than a page.
No. 05 · Step I
Matt Parker · 2014
Matt Parker used to teach mathematics in a secondary school and now performs it on stage and on YouTube, and his first book reads like one of his shows. Each chapter starts with something you can do at the table with paper, scissors, string or a pack of cards, and then follows the idea much further than you expected.
By the end he has got to higher dimensions, infinities of different sizes and how to build a working adding machine out of dominoes, all of it explained with diagrams and without assuming anything. It is less a puzzle collection than a set of experiments with questions attached. If you like making things, I have step-by-step pages for the missing square puzzle that go well with it.
Best forPeople who learn with scissors and string, and fans of his videos.
Step II · Books 06–11
Between about 1880 and 1930 a handful of men turned the newspaper puzzle into an art. Almost everything in a modern puzzle book is descended from these six.
No. 06 · Step II
Sam Loyd, selected and edited by Martin Gardner · 1914, this selection 1959
Sam Loyd was America's great puzzle maker, and he was also something of a rogue. He began as a composer of chess problems while still in his teens, moved on to advertising puzzles and newspaper columns, and for decades supplied the American public with brain-teasers dressed up in cartoons and tall stories. He claimed to have invented the famous sliding puzzle with fifteen numbered tiles, which historians now say he did not, though he did very well out of offering a thousand dollars for a solution he knew to be impossible.
After Loyd died in 1911 his son gathered thousands of his puzzles into a huge and chaotic Cyclopedia. Martin Gardner went through it, picked 117 of the best mathematical ones, fixed the errors in the answers and kept the original illustrations. The result is the most entertaining of the old collections. The puzzles themselves are sharper than the patter suggests, and several are still hard.
Best forReaders who like their puzzles with showmanship.
No. 07 · Step II
Henry Ernest Dudeney · 1907
Henry Dudeney was Loyd's English rival and, most people who know both would say, the better mathematician. He never went to university. He taught himself, worked as a clerk, and for some twenty years wrote the puzzle page of the Strand Magazine, the same magazine that was printing Sherlock Holmes. This was his first book, and its opening section borrows Chaucer's pilgrims on the road to Canterbury and has each of them set the others a problem.
The Haberdasher's is the famous one. He asks how to cut an equilateral triangle into four pieces that can be rearranged into a perfect square, and Dudeney's answer is so neat that he had it made as a hinged wooden model and showed it to the Royal Society. The rest of the book moves through a monastery, a castle and a puzzle club, with more than a hundred problems in all. The solutions are careful and often teach a general method.
Best forReaders who enjoy a puzzle wrapped in a story.
No. 08 · Step II
Henry Ernest Dudeney, edited by Martin Gardner · 1967
If you buy one book from this step, I would make it this one. Dudeney published two late collections, Modern Puzzles in 1926 and Puzzles and Curious Problems, which came out in 1931, the year after he died. Both went out of print and became hard to find, so in 1967 Martin Gardner combined them into a single volume, sorted the puzzles by type and added notes where later mathematicians had improved on Dudeney's answers.
The 536 puzzles are short, a paragraph each, and they cover everything he was good at: arithmetic and algebra in disguise, clock and speed problems, dissections, chessboard problems, matchsticks, and routes to be traced without lifting the pencil. Very few need more than school mathematics and a good many need real ingenuity. Gardner called Dudeney England's greatest inventor of puzzles, and here you can see why at five hundred different angles.
Best forAnyone who wants the largest helping of the best classical puzzle maker.
No. 09 · Step II
Lewis Carroll · 1885 and 1893
Lewis Carroll was the pen name of Charles Dodgson, who lectured in mathematics at Christ Church, Oxford, for most of his adult life. This volume binds together two of his books of puzzles, and they could hardly be more different. A Tangled Tale is a set of ten short comic stories, which he called knots, written for a magazine. Each one hides a couple of arithmetic or logic problems in the plot, and the appendix, where he marks his readers' attempts and scolds them by pseudonym, is as funny as the stories.
Pillow Problems is the austere one. It holds seventy-two problems in algebra, geometry and probability that Dodgson says he solved entirely in his head while lying awake at night, without writing anything down until morning. He gives the answers and then the mental working. They are harder than anything else in this step, and they show a side of the author of Alice that the stories only hint at.
Best forAdmirers of Alice who want to meet the mathematician who wrote it.
No. 10 · Step II
W. W. Rouse Ball and H. S. M. Coxeter · 1892, thirteenth edition 1987
This is the scholar's book of the classical period. Walter William Rouse Ball was a fellow of Trinity College, Cambridge, and a historian of mathematics, and in 1892 he set out to write down the recreations of his time properly, with their history and with the mathematics that explains them. The book went through edition after edition. From the eleventh, in 1939, it was revised by the geometer H. S. M. Coxeter, who added chapters of his own on polyhedra and kept it up to date for another half century.
It is organised by subject and not as a list of problems to solve. There are chapters on arithmetical tricks, geometrical fallacies, magic squares, the knight's tour and other chessboard questions, map colouring, mazes, the Tower of Hanoi, and codes and ciphers. You read it to understand why a puzzle works and who first solved it. Nearly every later writer on this list had it on the desk.
Best forReaders who want the theory behind the puzzles, in one reference volume.
No. 11 · Step II
David Wells · 1992
David Wells is not a Victorian, but his book is the best guide to them and to everyone who came before, so I have put it here. It contains 568 puzzles arranged by date. The first come from Egyptian papyri and Babylonian tablets. Then come the Greeks, a medieval collection made for the court of Charlemagne that holds the earliest river-crossing puzzles, the problems of Fibonacci, the Renaissance, and so on through Loyd and Dudeney to the twentieth century.
Reading it in order you watch puzzles travel. The same problem turns up in one country after another, centuries apart, each time in new clothes. Wells gives the source for every one and solutions for all of them in the second half of the book. It is the cheapest way I know to own the whole tradition, and the current reprint is published by Dover under a slightly shorter title.
Best forReaders who want four thousand years of puzzles in historical order.
Step III · Books 12–15
One man gets a step to himself. Gardner wrote the Mathematical Games column in Scientific American for a quarter of a century, and recreational mathematics as a field is largely his invention.
No. 12 · Step III
Martin Gardner · 1959, revised edition 2008
In December 1956 Scientific American printed an article by a freelance writer about hexaflexagons, strips of paper folded into hexagons that reveal hidden faces when you flex them. Readers all over New York started folding. The publisher asked the writer, Martin Gardner, whether he had enough material for a monthly column, and Gardner, who had studied philosophy and had taken no mathematics after high school, said yes and went out to buy every book on the subject he could find.
This is the first collection of those columns, in the edition he revised late in life with new material after each chapter. It has the hexaflexagons, the game of Hex, polyominoes, a chapter on Sam Loyd, probability paradoxes that still start arguments, and two sets of nine short problems. Gardner's gift was to write as a fellow learner, one step ahead of the reader. If the first chapter tempts you, I have a page on how to make a hexaflexagon.
Best forAnyone who wants to start Gardner at the beginning.
No. 13 · Step III
Martin Gardner · 1994
Every so often Gardner gave his column over to a batch of short problems, and those issues were the ones readers wrote in about most. For this slim book he went back through them and picked his own favourites, the puzzles with the briefest statements and the most surprising answers.
Some have become standards. There is the one about two cyclists riding toward each other while a fly shuttles between their handlebars, which has a hard solution and a ten-second one. There is the mutilated chessboard and the returning explorer who walks a mile south, a mile east and a mile north and finds himself where he started. None takes more than a page to state. It is the book to keep in a coat pocket, and it is one of the least expensive on this list.
Best forA cheap, pocket-sized sample of Gardner's short problems.
No. 14 · Step III
Martin Gardner · 1978 and 1982, combined edition 2006
These two books began as illustrated teaching material, and they keep the format: a short cartoon strip sets up each problem, and Gardner's commentary underneath takes it apart. The Mathematical Association of America now sells them bound together in one volume.
aha! Insight is about the puzzle that cannot be ground out. Each one looks as if it needs pages of algebra and then gives way to a single idea, and Gardner's point is that this flash, which he calls the aha reaction, is a skill you can practise. aha! Gotcha is its darker twin, a tour of paradoxes in logic, probability, numbers, geometry, time and statistics, from the liar who says he is lying to voting systems in which the public prefers A to B, B to C and C to A. For the most famous probability paradox of all, see the Monty Hall story.
Best forTeachers, and readers who like paradoxes more than calculations.
No. 15 · Step III
Martin Gardner · 2001
Gardner's columns fill fifteen books, and near the end of his life he was asked to choose the best of them for a single volume. He picked fifty, arranged them in twelve sections from arithmetic and geometry to infinity, games and physics, and wrote a new afterword for each one to say what had been discovered since. The book runs to about seven hundred pages and deserves its title.
The famous columns are all here. You get the Game of Life, which Gardner introduced to the world in 1970 and which I mention again in the biography of John Conway. You also get Penrose tiles, the paradox of the unexpected hanging and Newcomb's paradox. This is not a book of quick problems. Each chapter is an essay with puzzles inside it, and it will keep you busy for a year.
Best forThe one Gardner book to own if you are only going to own one.
Step IV · Books 16–19
Four books by Raymond Smullyan, a logician who was also a stage magician and a concert-level pianist. No arithmetic is required. Patience is.
No. 16 · Step IV
Raymond M. Smullyan · 1978
The book opens with a joke Smullyan's brother played on him at the age of six, and it does not stop being funny for two hundred pages. It also does not stop getting harder. The first puzzles are set on an island where knights always tell the truth and knaves always lie, and you are asked to work out who is who from a sentence or two. Then come Portia's caskets from The Merchant of Venice, an inspector sorting suspects, werewolves, and a Transylvania in which everyone is either human or vampire and either sane or insane.
All the while Smullyan is building something. The puzzles about people who make statements about their own truthfulness lead, step by step, to the paradoxes of self-reference, and the last chapter uses the same islands to explain Gödel's incompleteness theorem. There are more than two hundred puzzles. If the ending makes you curious about Gödel himself, his life is number thirteen on the biographies list.
Best forEveryone. It is the best book of logic puzzles ever written.
No. 17 · Step IV
Raymond M. Smullyan · 1982
The title comes from a short story of 1882 by Frank Stockton, in which a prisoner must choose between two doors, with a lady behind one and a tiger behind the other. Smullyan adds a sign to each door, tells you something about whether the signs are true, and lets you reason your way to safety. Twelve trials later there are nine doors and the signs have become thoroughly treacherous.
The first half of the book is a run of such puzzles, including a memorable asylum where you must work out whether the doctors or the patients are in charge. The second half is a short novel, The Mystery of the Monte Carlo Lock, about a safe whose combination can only be found by understanding a machine that operates on strings of symbols. By the end the reader has, almost without noticing, learned the ideas behind self-reproducing programs and undecidable problems. It is a little harder than its predecessor and just as good.
Best forReaders who finished number sixteen and want the sequel.
No. 18 · Step IV
Raymond M. Smullyan · 1985
This is the strangest of Smullyan's books and, for a certain kind of reader, the most rewarding. After some opening chapters of ordinary logic puzzles you enter a forest full of birds. When you call out the name of one bird to another, it answers with the name of a third. A mockingbird, hearing any bird's name, answers what that bird would answer to itself. From a few rules like this you are asked to prove things about what must live in the forest.
What you are really doing is combinatory logic, a system devised in the 1920s that turned out to be one of the foundations of computer science. Smullyan chose birds in honour of Haskell Curry, a founder of the subject and a keen birdwatcher, and each bird is one of the classical combinators under another name. People who program in functional languages treat the book as a rite of passage. It demands pencil and paper on every page.
Best forProgrammers, especially anyone curious about functional languages.
No. 19 · Step IV
Raymond M. Smullyan · 1979
In an ordinary chess problem you are asked what White should play next. In these fifty problems you are asked what has already happened. Holmes and Watson come upon a position on a board and Holmes deduces, from the placement of the pieces alone, which pawn made a capture, whether a king has moved, or on which square a missing piece was taken. Chess composers call this retrograde analysis, and Smullyan was one of its masters.
You need to know the rules of chess and nothing more, since skill at playing does not help. What the puzzles train is pure deduction, the elimination of everything that could not have happened until one history is left, which is exactly why Holmes is the right narrator. Watson tells the stories in a fair imitation of Conan Doyle. For another case where the detective's reasoning can be checked with mathematics, see Sherlock Holmes and the bicycle tracks.
Best forChess players, and fans of Holmes who know how the pieces move.
Step V · Books 20–23
Four books in which the puzzles start to turn into mathematics. Expect to spend an evening on a single page, and to enjoy it.
No. 20 · Step V
Peter Winkler · 2004
Peter Winkler spent years as a research mathematician at Bell Labs before moving to Dartmouth, and he collected puzzles the way other people collect wine. His rule for this book was strict. A puzzle had to be easy to state, need no advanced mathematics to solve, and still be capable of stopping a professional for a day. Many of the hundred or so he chose had been passed around by word of mouth among mathematicians for years.
So you get prisoners who must guess the colours of their own hats, ants walking on a stick, coins on a table, and lines of soldiers, each with a solution that is short, elementary and very hard to find. The final chapter lists puzzles that nobody has solved. Winkler's write-ups are models of clarity, and he always says where a puzzle came from when he knows. This is the book I would give to someone who has done the whole list above it.
Best forMathematics students and professionals who think they have seen every puzzle.
No. 21 · Step V
Frederick Mosteller · 1965
Frederick Mosteller founded the statistics department at Harvard, and this thin book is what he did for fun. The title undercounts, because there are actually fifty-six problems. Each is stated in a few lines and solved in a page or two, and the solutions are the real content, since Mosteller uses them to show how a statistician thinks.
The problems start with a sock drawer and go on through dice, duels, elections and broken sticks to some classics, among them the birthday problem, Buffon's needle and the prisoner who asks his guard which of the other two will be executed. Probability is the part of mathematics where intuition fails most reliably, and that is what makes these so satisfying. You will be confidently wrong several times. A first course in probability helps for the later ones, though the early problems need only fractions.
Best forStudents of probability and anyone preparing for a quantitative interview.
No. 22 · Step V
Anany Levitin and Maria Levitin · Oxford University Press · 2011
Many classic puzzles are really questions about procedure. How do you get everyone across the bridge in the least time, find the false coin in the fewest weighings, or move the tower of discs from one peg to another? Anany Levitin, who wrote a widely used textbook on the design of algorithms, noticed that such puzzles teach algorithmic thinking better than most programming exercises, and put 150 of them in one book with Maria Levitin.
It opens with a short tutorial on general strategies, such as working backward, looking for an invariant, and dividing a problem in half, each illustrated with a puzzle. The main collection is then split into easier, medium and harder sections, and every puzzle gets a hint, a full solution and a note on its history. No programming is involved at any point. Several of these have long been asked in interviews at technology companies, which is one reason the book sells.
Best forProgrammers and computer science students.
No. 23 · Step V
Dmitri Fomin, Sergey Genkin and Ilia Itenberg · American Mathematical Society · 1996
In the Soviet Union, children who liked mathematics went to a circle. It met after school, was usually run by university students, and worked through carefully chosen problems instead of following a textbook. This book is the working material of the circles of Leningrad, written by three people who taught in them and translated into English for the American Mathematical Society.
The first half is a year's programme for beginners, organised by idea: parity, the pigeonhole principle, divisibility, graphs, invariants, games. Each chapter opens with problems any twelve-year-old can understand and builds to ones that would trouble an undergraduate. The second half is a second year. It is the best book I know for showing how puzzles become mathematics, and it comes with notes for whoever is leading the session. The same system produced Grigori Perelman, and you can read about it from the inside in his biography.
Best forTeachers, parents and ambitious students from about twelve upward.
After the shelf
Puzzles are the door, and behind it is the habit of proving things. When you start wanting to know why a trick always works and not just that it does, you are ready for the best books for learning mathematical proofs, several of which read like puzzle books with the answers written out properly. The math books you can finish in one sitting are a gentler next step.
If the people interested you as much as the problems, Conway, Erdős and the Soviet circle system all have books in the best biographies of mathematicians. For younger solvers there is a whole shelf in the best children's math books. And if one grid of nine by nine is puzzle enough for you, I wrote about the mathematics of Sudoku separately.
Keep wandering
A few more pieces in the same spirit — math, design, and slow attention.

The 24 Best Biographies of Mathematicians
From Hypatia's Alexandria to a Moscow exam room in the 1980s, 24 lives of mathematicians, sorted by century and by how much mathematics each book asks of you.
Open
The 17 Best Books for Learning Mathematical Proofs
A reading order for anyone moving from computing answers to writing proofs, from the first textbook all the way to Gödel.
Open
How to Make a Hexaflexagon From a Strip of Paper
A strip of ten paper triangles folds into a hexagon with three faces instead of two. Pinch, push and open, and a face you never coloured blooms from inside.
Open
Sherlock Holmes and the Bicycle Tracks: Which Way Did It Go?
Holmes read a bicycle's direction from its tracks in the mud, and his reasoning was wrong. The right answer needs one tangent line. Try it on tracks you can test yourself.
Open
Marilyn vos Savant Was Right All Along
In 1990, Marilyn vos Savant answered a probability puzzle correctly. Nearly 10,000 people — many of them academics — wrote to tell her she was wrong. She was not.
Open
The Mathematics of Sudoku: How Many Grids, Why 17 Clues
How many Sudokus there are, why no puzzle gets below 17 clues, how computers and pencils solve them, and the 810-edge graph hiding in the grid.
Open