Many of the documents presented here have been OCRed, their characters having been automatically recognized by a software, then more or less proofread.
It gives a "searchable text" instead of a non-searchable image. Most OCRed text don't contain many errors of recognition, while others have more, due to the bad quality of the document we had to scan. For those bad ones, there is often an "image" copy at the end to check. Most official forms characters are barely legible: these parts have generally not been proofread since the form itself leads to the meaning of these uncorrected recognition errors.
The quantity of the documents and their searchability have been preferred rather than getting a perfect OCR: if really needed, one can obtain a scan of the original image before OCRing. (Write to the webmaster if needed.)
Some complex documents (with many frames etc) are presented partly OCRed, part image, since that solution was better rendering the original.
Some of the documents being impossible to OCR are presented in Adobe pdf image, not searchable.Adobe pdf formatting was preferred for most of the files. Some parts of some documents have been underlined (yellow or green) to catch the reader's attention, and stamps have sometimes been added to give some moreinformation.
The computer size of some large documents is given.