ahaspel

There's also a side-by-side option now. On any article, clicking the little scan button above the double-navigation arrows in the right margin will open the scan at whatever page you're viewing, and it will scroll as you scroll the text.

Thanks for the suggestion.

These were both pipeline errors and they have just been corrected, thanks to your sharp eyes.

This is now fixed, along with several more serious rendering errors in "United States". Thanks a lot for pointing it out.

Thanks for the kind words. I've had a few requests for a technical appendix (i.e., "how I built this") and it is in the works.

Not a bad idea. I'll see what I can work out on that score. But I imagine the far more common path is from the Guide to the encyclopedia than the reverse.

The Reader's Guide has been added to the ancillary material. Thanks for the excellent suggestion.

I wanted to let everyone know that article search from articles is now working properly again. A path problem. Apologies.

Thanks, nice catch. The tables can be tricky and I appreciate the heads-up on this markup leak. It will be corrected shortly.

It would indeed. I will see about working this in, it's highly pertinent.

I'm looking forward to it. The 9th is great in its own right and a lot of it is in the 11th. Alfred Newton's nearly 200 articles on bird species and a few classic essays by Macaulay come to mind offhand.

If you're reading an article, just go to the top and type in the left-hand search box. That will search for articles as well as text within articles. The right-hand box searches the text of the article you're reading.

I feel exactly the same way about encyclopedias and dictionaries. And Encarta really was amazing. You'd be surprised how much modern criticism of the 11th amounts to "no entry on the Great War", except in earnest.

I'm familiar with the Synopticon, which would be fun to structure.

I didn’t do OCR myself, except for the topic index and to fill in a few gaps. I started from existing Wikisource text and then built a pipeline around that: cleaning (headers, hyphenation, etc.), detecting article boundaries, reconstructing sections, and linking things back to the original page images. Most of the effort went into rendering the complex layouts, and handling the cross-linking, not the initial ingestion.

Glad to go into more detail if you’re interested, but that’s the gist of it.

Under the hood it’s not XML-TEI — it’s a relational/data-pipeline approach, with article boundaries, sections, contributors, cross-references, and source-page provenance all reconstructed into structured records. The text itself is public domain, but I haven’t released a bulk structured export yet.

People asking for dataset access has definitely been one of the themes of this thread. I’m taking that seriously. If I do expose it, I’d want to do it in a form that preserves the structure and doesn't just dump plain text.