What is this site all about?
This site celebrates NYT crosswords and the people who make them.
It provides a comprehensive list of every daily puzzle and most Variety puzzles
going back to 1967.
Puzzles with special attributes are noted. Records in various categories are tracked.
In particular, Shortz-era data is organized by constructor — a feature unique to this site.
See the Grand Tour for a list of features organized by target audience.
What can I do here?
The best approach is to poke around and explore. You can view lists of puzzles that meet
various criteria, click on a puzzle to see it displayed in full with the answers, click on
an answer to see what clues have been used for it, click on the displayed date to see the
answer word in context, etc.
Why is registration required?
Your donation dollar goes largely to running the site or to
Treehouse for Kids
, a Seattle non-profit providing education
assistance and mentoring to foster children.
What parts of XWord Info can I access for free?
If all you are looking for are answers to recent clues, the 45 most recent puzzles are
always freely available.
Pre-Shortz puzzles are also always available. In fact, you can both see them
and even solve them online for free!
Most other pages require you to sign up for an account.
Can I link to XWord Info pages from my blog?
I'm glad you asked! I've added some logic to allow pages requested via links on blogs and a few other sites to
come through without registration.
If links from your blog or site don't seem to get through, send me mail and I'll update that logic.
Who is this site for?
Mostly for me, but crossword constructors will find it useful and enthusiasts will
find it interesting.
When I started blogging about crosswords, I wondered what I could learn about NYT puzzles if I ran some statistical analyses
on the puzzles themselves.
In October 2008, my blog moved to The New York Times where I wrote for over two years.
I have since retired from there but I still sometimes comment about puzzles on
my personal blog.
Is the data here 100% accurate?
I do my best to expand rebus entries (including ones that are interpreted differently in different directions)
and even correctly parse answers that go around corners or do other such tricks, but there are likely
still bugs. Please let me know if you spot any.
Where does all this crossword data come from?
Puzzles from October 23, 1996 to the present come from the Across Lite files on the nytimes.com website.
Originally, that was all I thought I would ever be able to get but it turns out Barry Haldiman has been
collecting crosswords in electronic form for a long time and he has graciously offered to share some of
these with me.
Thanks to Mr. Haldiman, this site now includes NYT puzzles covering the entire Will Shortz era, i.e. going back
to November 21, 1993. Mr. Haldiman's database is available online
and it has all kinds of insight and analysis in areas I don't cover.
In 2012, David Steinberg started The Pre-Shortzian Puzzle Project
dedicated to digitizing earlier NYT crosswords.
As those older crosswords become avaialble, they get added here.
Pre-Shortz stats are kept separate from Shortz-era data.
Constructor photos are courtesy of the puzzle authors themselves. I'm grateful to them all.
What do FITB, XRef, and all the various other stats mean?
- Row and Column count refers to the size of the grid
- Words is the number of clues or answers. Single answers that are phrases count as one word.
NYT guidelines call for a maximum of 78 words for a 15x15 puzzles, and 140 words for a 21x21 grid.
These are sometimes exceeded to handle special themes.
- Blocks is the number of black squares.
- Average word length is an obvious mathematical calculation. Puzzles with higher averages
tend to be harder to construct.
- Missing letters are ones that are absent from the grid. Puzzles that use all 26 letters
are said to be pangrams.
- Scrabble score lists both the total tile value of all the letters as well as the total divided by
the number of letters.
- Open squares is an XWord Info invention. Grids seem open if they have large areas of white
space. This calculation counts the number of white squares that are not next to black squares, even
- Cheater squares are black squares that can be removed without affecting the overall word
count of the grid. Constructors usually try to avoid these but some are often inevitable and many
great puzzles have plenty of cheaters. If you click the "Analyze" button, you'll see cheater
squares marked with a + in the colorized grid.
- FITB is the number of Fill In The Blank clues.
- XRef is the number clues that cross-reference other clues, such as "see 14 Across".
- Debuts is the number of words that first appeared in this puzzle. They are colored differently
in the answer section. Sometimes the number of words that are reused (they debuted on this puzzle but
were picked up by other constructors later) is also shown.
- Spans counts the number of answer words that span the full grid. Double, triple,
or quadruple stacks are also noted.
- Freshness Factor is an attempt to algorithmically determine how
"fresh" a grid is based on how many times we've seen its answer words used before.
You can see details of the calculation here.
What is the XPF link shown on some puzzles?
This is for programmers only. Data for recent puzzles can be displayed in
. There is also a JSON format
Why are those stats sometimes highlighted?
If they are statistically significant in some way, perhaps there is a particularly low block count or high Scrabble average,
they are highlighted in a yellow block.
Click the highlighted link to see the relevant stats page and it should be clear why the stat is called out.
Why don't the dates here match the dates in my newspaper?
Dates here match the publication dates in the New York Times.
In syndication, the Sunday puzzle runs a week behind, and the dailies are five weeks behind.
There is a link to the syndicated puzzle on the home page, or you can go to the
page and count back the appropriate number of weeks.
To improve browsing speed, those mini-grids are cached locally using HTML 5 local storage, a technology only
available in modern browsers. Upgrade to the latest versions of Internet Explorer, Firefox, Safari or Chrome to see them.
Tooltips on hover are not available on touch devices like iPad.
How did you come up with the cool visual design?
I didn't. Site design is by Robin Troy.
Are you the same Jim Horne who wrote the original Microsoft version of the solitaire card game FreeCell?
What technologies were used to build this site?
XWord Info is built almost entirely on Microsoft technology. It is an ASP.NET 4.5 application
running on the IIS 8.5 Windows web server.
LINQ is used to manage the data stored in SQL Server 2014.
Software development is done on Visual Studio 2013.
Server-side code is written in C#.
All pages declare the HTML 5 doctype. Now you know.
Where does the dictionary of words not found in NYT puzzles come from?
It comes from Princeton University. In order to make it available on XWord Info, I'm
required to provide this notice:
WordNet 3.0 Copyright 2006 by Princeton University.
All rights reserved.
THIS SOFTWARE AND DATABASE IS PROVIDED "AS IS" AND PRINCETON UNIVERSITY MAKES NO REPRESENTATIONS OR WARRANTIES,
EXPRESS OR IMPLIED. BY WAY OF EXAMPLE, BUT NOT LIMITATION, PRINCETON UNIVERSITY MAKES NO REPRESENTATIONS OR WARRANTIES OF
MERCHANT-ABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OR THAT THE USE OF THE LICENSED SOFTWARE, DATABASE OR DOCUMENTATION
WILL NOT INFRINGE ANY THIRD PARTY PATENTS, COPYRIGHTS, TRADEMARKS OR OTHER RIGHTS.
This useful word list includes many proper nouns, names, phrases, etc.
Even after removing all items shorter than 3 letters, longer than 21 letters, and any containing digits like Y2K,
there are still 142,539 entries.
XWord Info is © 2007-2014 by Jim Horne. Crossword puzzles are © 1967-2014 by The New York Times.