Marc Andreessen proposed the HTML IMG tag on a mailing list on Feb. 25, 1993, and Tim Berners-Lee wasn't sold

It is easy to talk about the IMG tag as if it arrived from nowhere. It didn’t. It arrived in an email, late on a Thursday night. The date header on message 174 of the www-talk archive reads Thu, 25 Feb 93 21:09 -0800, from marca@ncsa.uiuc.edu, subject “proposed new tag: IMG.” It opens: “I’d like to propose a new, optional HTML tag:”. Then, a few lines later, the line that does the heavy lifting: “This is required functionality for X Mosaic; we have this working, and we’ll at least be using it internally.”

That second sentence is the real argument. It is a proposal that says the thing already exists. You can almost hear the shrug.

The first reply came the same night. Tony Johnson, writing at 11 p.m. Pacific from SLAC, said (message 175): “I have something very similar in Midas 2.0 (in use here at SLAC, and due for public release any week now).” His version used an ICON tag with a name and an href, plus a separate INCLUDE idea. So the web had two browsers, and each already had its own way of doing pictures.

Then came CERN. Tim Berners-Lee replied the next day (message 178) with a different route: an anchor with a REL value of “EMBED, PRESENT”, so that existing markup could carry the image. His line: “But I hadn’t wanted a special tag.” Later that day, in a reply to someone else’s question, he wrote: “I don’t want to change HTML now if I can help it, until it has gone to RFC track.” It is the voice of someone protecting a spec while the browsers sprint ahead of it.

Jay C. Weber came at it from yet another side (message 184): “I don’t think we should add idiosyncratic hooks for media one at a time.” He suggested MIME typing. Andreessen agreed MIME mattered, writing in his reply to Johnson that his tag “isn’t a substitute for the upcoming use of MIME as a standard document mechanism.” In other words, the person who proposed the tag already thought it was a stopgap. (It has been a stopgap for thirty-three years.)

One small exchange is my favorite. Jim Davis said he liked the functionality and was “indifferent to the syntax, so long as we get a standard,” then asked why the tag used SRC instead of HREF. Andreessen’s answer, in message 188: “Because I wanted to avoid overloading HREF – doesn’t really make sense in this context, I don’t think.” That is a design decision made in a single sentence, in a reply to a stranger, and every web page since has had to live with it.

What I can’t tell you from these pages is how the argument ended. The thread I read ends with Andreessen’s replies and no decision, and other accounts of what happened next disagree on dates and details, so I’m leaving them out. What the archive does show is a web being built in public by people who disagreed about nearly everything and wrote it down anyway. There was no vote and no spec meeting. A working browser and a short email did the persuading.

The messages in between 174 and 188 are a mixed bag. Some are about the IMG tag, and others are about X window memory, an httpd patch and search escaping. That is how real standards history looks: the famous bit sits between a bug report and someone’s cron job.

Nothing in the thread settles it.