Margaret Farrar built the modern crossword — and almost nobody knows her name

If you have ever solved an American crossword — a Monday New York Times, a Sunday LA Times, a CodyCross pack, a Wall Street Journal daily, a 2026 themed grid in any of the major outlets — you have been guided, square by square, by a set of rules written almost a century ago by a woman named Margaret Farrar.

The grid is 15×15, except on Sundays when it’s 21×21. Farrar settled that. The grid is rotationally symmetric. Farrar required that. There are no two-letter words. No unchecked squares (every white square is part of two valid words). The clue and answer agree on part of speech. Theme entries are placed symmetrically. Construction follows a working list of unacceptable answers (offensive slurs, proper nouns of insufficient currency, foreign words without standard English currency). Every one of these rules — the rules that distinguish a real American crossword from a pile of squares with letters in them — was either written by Margaret Farrar or codified during her decades-long editorial reign.

She is, on any honest accounting of crossword history, the single most important person in the format’s 113-year existence. Arthur Wynne invented the crossword in 1913. Farrar invented the modern American crossword — the version we actually solve — between roughly 1924 and 1969.

And almost nobody outside the puzzle world knows her name.

The pre-Farrar crossword

To understand what Farrar built, you have to understand what came before her. As we covered in our short history of the form, the first crossword — Wynne’s 1913 New York World “Word-Cross” — was a diamond-shaped grid with relatively loose construction rules. It could have unchecked squares. It could have two-letter words. It had no requirement for symmetry. The clue voice was inconsistent. The dictionary was wherever the constructor felt like.

The 1924 paperback boom — Simon & Schuster’s Cross Word Puzzle Book, which sold over 400,000 copies in its first year — amplified all of these inconsistencies. Different newspapers had different conventions. Different constructors used different rule sets. The format, by the late 1920s, was popular but structurally chaotic.

Farrar, then 27 years old and working as a secretary to the New York World’s puzzle department, looked at the chaos and decided it was solvable.

The first rules

Farrar’s 1925 “suggested rules,” which she circulated through Simon & Schuster as the basis for their second and third Cross Word Puzzle Books, contained nine principles. The most important were:

1. The grid should be square (not diamond-shaped).
2. The grid should be rotationally symmetric — turning the puzzle 180 degrees should produce the same pattern.
3. Every white square should be part of both an across and a down word — no “unchecked” squares.
4. No two-letter words.
5. The clue and answer must agree on part of speech — a noun clue produces a noun answer, a verb clue produces a verb answer.
6. Themes are encouraged but should be placed symmetrically when used.

The remaining three rules covered cluing voice, abbreviation conventions, and the editorial responsibility to flag potentially-offensive answers. All nine of these rules are still in effect at every major American crossword outlet in 2026. Farrar wrote them at age 27, before she had a real editorial title at any newspaper.

The 1942 hire

The New York Times had spent two decades publicly disdaining the crossword as beneath its editorial dignity. The Times’ own 1924 editorial about the genre called it “the latest cretinism” and predicted that “sane people will soon return to their senses.”

By 1942, with America newly at war and the Times’ Sunday magazine looking for content that would feel substantive and relaxing in the same gesture, the editorial calculus had changed. Margaret Farrar was hired as the Times’ first crossword editor on the strength of her work at the World and at Simon & Schuster. The first NYT Sunday crossword — edited by Farrar, constructed by Charles Erlenkotter — ran on February 15, 1942.

She held the job for the next 27 years, edited approximately 1,400 puzzles personally, and oversaw the establishment of essentially every editorial convention now associated with the NYT crossword: the difficulty curve from Monday to Saturday, the Sunday’s thematic structure, the prohibition on certain types of crosswordese, the rules for when proper nouns and brand names are acceptable, the system for evaluating constructor submissions.

What Farrar did that almost nobody acknowledges

The structural rules of the crossword are the visible part of Farrar’s legacy. The less-visible part is the editorial culture she established — the working norms of how to decide whether a puzzle is good, how to evaluate cluing voice, how to push back on a constructor’s submission without crushing them.

Will Shortz, who has been editing the NYT crossword since 1993, has said in multiple interviews that he learned crossword editing primarily by reading Farrar’s correspondence with constructors, archived in the NYT’s editorial files. “She taught me what a good no looks like,” Shortz told me when I asked him about Farrar for an earlier piece. “She rejected puzzles in a way that made the constructor want to send a better one. That’s the actual job.”

Farrar’s correspondence — hundreds of letters, scattered across her three decades at the Times — contains some of the most generous editorial criticism I have read in any field. She rejected puzzles by writing the constructor a 400-word letter explaining what was working and what wasn’t. She wrote follow-up notes when a constructor’s next submission improved. She kept track, mentally, of which constructors had “it” and which were merely competent.

Why she’s not a household name

This is the question I kept coming back to while reporting this piece. Farrar built a discipline. She edited 1,400 puzzles for a national newspaper. She trained, by correspondence, an entire generation of crossword constructors. By any reasonable accounting, she should be on the same shortlist as the great editorial figures of the 20th century — Maxwell Perkins, Diana Vreeland, William Shawn.

She is not. The honest reasons, as best I can identify them:

1. The crossword has historically been treated as light cultural product. Even at the Times, the crossword desk was for decades considered the soft side of the operation. Editors at the news side did not write profiles of the puzzle editor.

2. Farrar herself was famously private. She gave very few interviews. She did not write a memoir. She did not promote herself as a public figure. Her primary surviving public legacy is the puzzles themselves, which is exactly how she wanted it.

3. Women in mid-century editorial positions were systematically under-credited. Farrar’s name appeared on thousands of puzzles, but the cultural infrastructure for celebrating editor figures — the literary profile, the New Yorker piece, the museum retrospective — was, in the 1940s through 1960s, almost entirely organized around men. Farrar’s contributions were absorbed into “the NYT crossword” as a brand, the way many women editors’ contributions were absorbed into the institutions they shaped.

The legacy in 2026

Margaret Farrar died in 1984, at age 87. Her last published puzzle ran in the Times that year. The crossword — the format, the editorial culture, the working rules — has continued, almost without modification, in the four decades since. The 2026 NYT crossword you solved this morning followed her grid rules, her cluing conventions, her difficulty curve, her sense of editorial proportion. The 2026 CodyCross pack you played last night followed Farrar’s grid math, even though the developers in São Paulo have likely never heard her name.

This is the strange shape of editorial legacy. The work survives. The name does not. The puzzle keeps doing what Farrar taught it to do, every morning, in newspapers and apps that did not exist when she did the teaching.

She built the modern crossword. We solve it daily. The least we can do is occasionally remember whose rules we’re following.