It was common to make tables and use them to assemble a bitmap, where each cell had zero border/margin/padding and an exact size, and contained a "slice" of the image. Web authoring tools (and Photoshop) even had explicit support for generating this sort of thing, as I recall. This was I guess simpler to automate than defining clickable regions of a single image, and it allowed for the individual pieces of the image to be requested in parallel on slow connections (adding another dimension of progressive loading).
Looking at the code, it has definitely been modified from the original... there is now CSS as well as a google ad tracker... but visually it's probably almost exactly the same.
devin|4 months ago
zahlman|4 months ago
ranger_danger|4 months ago
hosh|4 months ago