I’m using a small js to fix this:

var img = new Image(‘logo_white.png’);

This way the browser will download the image and correctly display it on the print version.

A CSS border wouldn’t really solve the dark logo on white backround issue, so we’d still need two logo’s regardless of method so it’s not really causing us more work to maintain this in the long run.

For most situations I’d agree that this might not be suitable for everyone, but hey, it might be useful for someone

Using two methods like this also leads to much confusion if the logo is ever updated at a later date.

I would personally just put a css border round your white for the print.css I think borders still print out.

As for the print stylesheets, I tend to use either image replacement on the h1 tag or add an image inside the h1/div#logo.
I also add a hidden div bellow that with the printable version of the logo, usually black and white.
So when you print it you’ll get the title of the page (that comes with the alt of the image or the text within the h1) and then the logo.

I guess it depends on how you think your website is going to be used when printed, but this is usually good enough for my projects

Because our logo also contains the “University of Leeds” title I chose not to include a header tag also.

I’d think as long as the logo has the neccesary alt tags and title attributes then this would be enough meaningful markup, our pages have title attributes along the line of “Page title – University of Leeds” and are chock full of metadata, both general and Dublin Core

I wanted to avoid a situation where if CSS was disabled, then the user would get the logo (which has the UoL text) and a header tag repeating the UoL text).

The issue I see with this is that you have made your mark-up less meaningful. Striping out the header tag and the ‘University of Leeds’ text. OR am I missing the point and you are going to use both the H2 image replacement and the hidden image?