⌂ Puzzle Tools

Word Search Generator

Consumes a themed word list, produces a batch of recipe-stamped puzzle + solution pages as one run ZIP — cached for Book Builder, downloadable for checking. Every list word must place or the batch fails loudly.

Theme list

No list loaded.

Print size

Pick a print size. Grid, word count and letter size are set together — the point size shown is what actually prints at 8.5×11.

Puzzle settings

Costs no type size — Helvetica Bold shares most of its uppercase widths with Regular, and the widest bank entry grows about 1.7%. Bolding the GRID darkens the filler too, and in a word search most of the grid is filler. Solutions are unaffected: the capsule outline is what does the work on that page.
The bank's type is set by its WIDEST entry against one column's width, book-wide, so a single long word sets the size for every page. Auto takes the widest layout that stays under eight rows, which keeps the bank short and the grid tall — and keeps the type small. Two columns doubles the width per entry: measured at 18 words on an 18x16 grid, the bank goes 14.4pt to 17.3pt. It costs rows, and rows cost grid. At 14 words both fit (20.0pt grid, 17.3pt bank); at 16 the bank needs eight rows and the grid drops to 19.2pt. Fewer words alone changes nothing — it is fewer words letting you pick 2.
The default was labelled “match the print size” and does not: it is a fixed 17-unit nominal that works out to 17.3 pt in every book, which is 86% of the grid letter at 20 pt and 60% at 29 pt. It is the ceiling, not a result — at 29 pt a three-column bank already reaches it, so switching to two columns there buys nothing. The sizes above 17 pt exist to break that ceiling in a large-print book, and they cost grid height: measured at 29 pt on a 12-column grid with 9 words in 2 columns, 24 pt gives a 23.8 pt bank but clamps a 12-row grid to 24.5 pt. Drop to 11 rows for 26.6 pt grid against 23.8 pt bank, or 10 rows to keep the full 28.8 pt — but at 10 rows an 11-letter word can only run horizontally. A left bank costs grid WIDTH instead, and the width is set by this; dropping from 13 pt to 9 pt buys about two more grid columns. The bank is read, not solved.
Only applies to a left bank. The grid is where the solver writes, so keeping it on the outside puts the circling away from the spine curve; the bank is only read.
How many puzzles to make. Fill list makes as many as the list holds without repeating a word — the number cannot be worked out in advance, so it is found by running. A slider is hopeless at this range on a phone.
Higher = more shared letters between words (harder to spot).
Grouped lets Book Builder start a new section at each sub-theme. As dealt is for a series volume: the dealer already spaced the themes and checked the spacing, so any other order throws that work away.
A sub-theme normally spends its remainder on one short closing page rather than stranding it — a bank of 49 at 10 a page prints four full pages and one of nine. Tick this where the list has banks to spare and an uneven page is not worth the words saved. Leave it OFF for a place-name book: those closing pages are what print the 484 town names the floor exists to save.
House rule is one. A bank that cannot be filled within the cap fails loudly rather than quietly printing a puzzle with no single words in it. Raise it only for a list that is multi-word by nature — hymn titles, film titles — where the entries ARE the theme. 99 is off.
Off draws a rule between the grid and the word bank instead. The border and the rule do the same job — separating the letters from the list — so a page carries one or the other, never both. A facing-page bank has nothing to divide, so it gets neither.
Draws in familiarity order where the list carries a familiarity score, and in level order — common, then familiar, then specialist — where it does not, so whatever the batch cannot use is the least-known end of the list rather than an arbitrary tail. It also front-loads each sub-theme: its first puzzles hold its best-known words and its last ones its most obscure. A score is a weight on the draw and never a filter: no word is held back for scoring low.
Every word is used at most once across all puzzles. Sub-themes are retired once they cannot fill another puzzle, rather than being drawn again. Fails loudly if the list runs dry before the batch is done.
Sweeping pools the words each retired sub-theme left behind into puzzles titled “Mixed”. Right for a hodgepodge book; leave it off for a single-theme book, where every puzzle should carry a real sub-theme name.
The repeat box narrows “once per book” to “once per sub-theme”: Andover is a real place in twelve states, and with states in the puzzle headings each appearance is a different fact. Right for the town books; leave it off for vocabulary, where the same word under two sub-themes is just a repeat. Mixed puzzles have no heading to tell two copies apart, so they never take a name the book has already spent.
Load a theme list to begin.
b0915-0001