[Binchen]

Originally Posted by Tex2002ans

An image of a table is worst:

<img alt="" src="../Table1.png"/>

Adding an alt tag is better:

<img alt="Table 1" src="../Table1.png"/>

I agree on all points, but the example above is not a good one. The alt text should contain something meaningful when the image is not displayed, which is the case with text-to-speech. Here a good software would use the alt-text, and "Table 1" is now as meaningless as nothing. In the last case you can always see that there is something, but that's it.

You can see it especially well in initials, where the first letter of the chapter is represented by a picture. I have seen it many times: the alt text then reads "Letter A" (assuming the image shows letter A). This is nonsense. The correct alt text is "A", not "Letter A" or "Initial A" or anything else. If I have the publisher's logo on the cover, then the correct alt text is "Raqndom House", "Penguin Books" or whatever the publisher is called, but neither the name of the image "publisher-logo.png" or "logo" or anything like that makes sense.

In your case "The e-book creator did not care about disabled people" should be in the title attribute or further hints in the longdesc attribute, where in the longdesc attribute only URIs may occur. EReaders certainly don't evaluate this, whether reading software does it I don't know.

For descriptive content the alt-attribute is the wrong place.