Fold3: Military records online

Fold3 Blog

The official blog of Fold3

Adding more horsepower to Footnote search

Under the hood of Footnote lies a powerful search engine that keeps track of everything – nearly 300 million images, documents, annotation, comments and stories.

But with the tremendous growth in search traffic on the site, we realized months ago that we needed more horsepower.

For example, our 1930 US Census project alone will add 120 million new people and more than 3 million images.

Starting early this week, we began upgrading our search engine. But it’s a feat akin to trying to swap out your 4-cylinder engine with a V8 while driving down the highway at 65 mph.

As a result, some of you noticed some sluggishness in searches – some bumps in the road.

The good news is that the first phase went live today without a problem (20:05 GMT).

Although, we’re still cruising down the highway, it will be a few more days before we get firing on all cylinders.

Once that’s complete, what does all this effort this mean for you? You should see:

• Quicker search results.
• New content and Member contributions will be findable almost as soon as they are added.
• More historical content added to Footnote will not affect search performance.

We know there will probably be a few hiccups as we fine tune this new engine and apologize in advance for any inconvenience.

Please continue to kick the tires and let us know of any issues you find.



  1. I’ve noticed that the search engine only picks up 50% of the info in a document. It doesn’t even pick up all the personal names and basically no place names unless an individual user has annotated the place. Is there a way to search every word in the documents? Also, is there a way to download a file as a batch? For instance some of the Rev War Pension files have 20-30 pages. Instead of having to download each page individually, can you select the whole file and download it all at once?
    Thanks for your time and help?

  2. In the case of hand written documents, the search engine will only pick up information that has been typed in somehow–this would include annotations on the image that are added by Footnote or by members and names, places, dates that have been keyed from the image, but don’t show up as annotations.

    Most typed documents on the site have been run through an Optical Character Recognition (OCR) process which does a pretty good job as long as the image is clean, but often makes mistakes and doesn’t know if a word is a name, place, etc.

    On batch download, we’re still thinking about possible solutions for this. With the size of the images on the site, batch downloading can quickly become a real bandwidth hog. It sure would be a nice convenience, so hopefully we can come up with a good solution.

  3. trying to trace descendents of Gillies Hercus who lived in St Louis Missouri at least from 1894-1917