Crossword puzzle maker for printable grids

How many words actually fit in a 15 by 15 grid?

Measured against a list of common everyday words: 20 words in, 19 placed. Thirty words in, about 24 placed. Forty words in, still 26. Past roughly 25 placed entries a 15 by 15 field gets tight, and each further word adds almost nothing. Throw 100 words at it and you get somewhere between 24 and 31, depending on word length, because the rest finds no free crossing. Short words built from common letters pack in far better than long subject vocabulary. The other sizes scale the way you would expect: 13 by 13 holds around 20 words, 17 by 17 around 32, 21 by 21 around 50.

That number is the reason for the placement line above the grid. Plenty of makers advertise “6 to 30 words” and quietly drop nine of them at 30. You get a good-looking grid and only notice at proofreading time that three of the vocabulary items you needed are missing. Here the count sits at the top, with a list of every word left out and the reason for it.

When the reason is lack of room, a bigger grid helps and the button is right beside the list. When the reason is “no letter in common”, a bigger grid changes nothing. That word simply has no overlap with the rest of your list, and the fix is a different word or a longer list.

How should you write clues that are worth solving?

The grid is the cheap part. A generator arranges 30 words in under a tenth of a second; writing 30 clues that are fair, unambiguous and pitched at the right audience is the work. Three habits carry most of the weight.

Match the difficulty to the reader, not to yourself. A clue that names the category and one distinguishing property suits a classroom sheet: “Plant with trunk and crown” for TREE. A party puzzle can lean on shared memory instead: “Where Anna dropped the cake, 2019”.

Keep the part of speech consistent between clue and answer. If the answer is a plural, the clue should be plural. Solvers use that agreement to check a guess before they commit ink.

And keep at least one clue easy enough to be a way in. A puzzle where every entry is hard is not harder, it is just unsolvable, because interlocking grids are solved from the crossings outward.

What is the difference between a crossword and an arroword?

In a numbered crossword every entry carries a number and the clues sit in two lists beside the grid, one for across and one for down. In an arroword, the style common in Scandinavian and German magazines, the clue is printed inside a blocked square in the grid itself and an arrow shows which way the answer runs. No numbers, no clue list. That is a different data model rather than a different skin, and this tool builds the numbered form.

Worth knowing if you set puzzles in German: the phrase “amerikanisches Kreuzworträtsel” in German puzzle usage means a grid with no preset black squares, which is exactly the criss-cross style built here. It has nothing to do with the American construction school and its symmetric black-square patterns.

Why does the seed matter?

Each grid is the best of 200 attempts with different word orderings. So that the same list does not produce a different puzzle every time you reload, every attempt hangs off a number, the seed, and it is shown next to the layout controls. Note the seed with your word list and the same grid comes back next term, square for square.

That sounds like a detail for programmers, and it is the difference between having a worksheet and having a template. Print 24 copies, come back three weeks later needing four more, and you want the identical grid rather than a fresh one. Most makers offer nothing but a reshuffle button.

How do you get a clean A4 print?

The print window carries its own document with no site header, navigation or footer. The page is fixed to A4 portrait with a 15 millimetre margin, and every measurement in the grid is given in millimetres. That is not fussiness: viewport units resolve against a substitute viewport when printing and land differently in different browsers, which is how the same puzzle arrives at a colleague with its left edge cut off.

Colour carries no meaning in the output. Squares are defined by their borders, and hidden-word squares by a doubled border plus a small index digit. The reason is in the specification for print colour adjustment: a page asking for background colours to be printed is making a suggestion that the browser and the user setting are free to overrule. A puzzle whose squares exist only as background fill arrives as a blank sheet whenever background graphics are switched off.

Two things a web page genuinely cannot control, and both belong in an honest answer. The browser header and footer with the address, date and page number is a checkbox in the print dialog. Whether your printer can go borderless is up to the printer.

Can the puzzle be solved on screen?

Yes, and the grid follows the ARIA grid pattern so a screen reader can work with it. Arrow keys move between squares, Home and End jump to the start and end of a row, Ctrl with them jumps to the start and end of the grid, and the space bar switches between typing across and typing down. Blocked squares are announced as blocked rather than skipped, so the reader keeps its bearings in the raster.

There is deliberately no real input field inside the squares. One input per square makes the common screen readers announce twice and switch modes on every letter. The grid captures keystrokes itself instead, and every clue in the list has an ordinary text field next to it as a second, equally complete route. Solvers who think in words rather than squares use the list.

The share link packs the word list, clues, grid size, umlaut setting, hidden word and seed into the fragment after the hash. A 30-word puzzle comes out at roughly 350 to 400 characters, depending on how long the clues are. Browsers never send that part of an address to a server, so a shared puzzle stays between the two browsers involved.