It might be prudent to step back for a second and think about what is occurring with this technique. When we are importing the html file, the html file itself is read as a string and the image is not rendered. But once loaded into the document, the browser then interprets the html file as an html file and consequently the image does indeed render. The xmlhttprequest object retrieves the document; the html markup renders it, just like in the earlier xml example with a namespace. If you keep the head section of the imported document relatively free, then we don't hit the one connection snag and instead are using multiple connections constantly.

There is also a conceptual framework that needs to be explained in terms of using ajax effectively. Once the html file has been loaded it would be erroneous to think of that html file as a separate window instance or a separate frame or even a separate document.