You can’t quote me on it!


This is the latest installment in a regular series of researched and reported stories that @dave asked me to write about things of broad interest to the community. Previous articles have tracked down the history of SHOUTY CAPS, the first instance of “this page intentionally left blank,” and the technological hope bundled in the Voyager probes.

The mark we know as a greater-than sign or a right-pointing angle bracket has a lot of modern meanings. One of its most ancient, however, was as a quotation mark. Manuscripts starting over two millennia ago show a > — a diple — before each line of text that’s referenced from another source.

That use persisted for centuries, until it was replaced in Roman and other alphabetic texts with quotation marks, often in pairs. Hundreds of years passed, and then the > sign rose again as a quotation symbol in margins, this time in Usenet newsgroups. But why?

I consulted people whose experience with the Internet dates back to some of the earliest internetworking of nodes, including Dave Crocker, an early contributor to the evolution and standardization of email, and many people involved in developing and fostering early Usenet newsreading software.

With their help, I’ve pinpointed what appears to be an inflection point, but nobody seems to recollect, nor can we find documentation, of how it arose. There’s a point between 1979 and 1983 at which everybody suddenly knew that a > meant a quote from a previous message.

But no one knows why it happened.

Quotation as a modern concept

I recently looked into curly quotes for the Atlantic, wondering if their demise was at hand. That led me to punctuation-mark historian Keith Houston, whose research finds Samuel Richardson, the inventor of the English novel and other innovations, likely first used what we think of as an intra-line starting and ending set of quotation marks in 1748.

But he said that prior to Richardson, quotes in books were mostly to mark excerpts from letters and sometimes other sources, not for dialog, which was typically inferred. These kinds of insertions have been marked in books and manuscripts back to the third century b.c.e., he said, and used something that looked a lot like a > not long after such marks first appeared. (Houston wrote a lot more about the quotation mark in this Slate article, and about punctuation of all kinds in his wonderful book, Shady Characters.)

I wondered if there were a connection to the modern use: could a classicist-turned-computer-scientist have thought of the diple and plopped it in for quoted material? Such a person, if they existed, didn’t document that thinking when the > came in use in the late 1970s or early 1980s.

Before the bracket was used for quoting previous messages — or other content not written by the poster or emailer — came the indent. This seemed a little backwards to me, as indenting wrapped text requires a more sophisticated email or news-reading program than existed in the 1980s. But in those days, all text was hard wrapped, with a line return at the end, before it was transmitted. Indents were just an algorithmic insertions of tabs (or hard-coded spaces) on wrapped lines handled by email or news composition software.

At some point, netnews software began using the > instead of indents, and nobody is quite sure why. David Wright posted a message on June 29, 1983, in which he mentioned the convention as being obviously well established:

Somewhat more acceptable is the practice of starting each quoted line with a ‘>‘. Unfortunately, it looks kinda ugly.

And many people I spoke with who were writing fundamental code and running parts of Usenet around and before that time recall it already being in use. But none could find the spark that make it prevalent.

Through hundreds of searches on Google Groups, which contains the largest extent archives of Usenet dating back to its earliest days, I found this unremarkable October 11, 1983, message — the contents of which still resonate to this day — in which this exchange appears:

	In response to:

	>    Congratulations, girls!
	>    You’ve come a long way, baby.

	Oh, dear.  Apparently not far enough.  I am quite a few years beyond
	being a “girl,” and quite a few more beyond being a “baby.”

That appears to be the first use in text of > to quote a message. (I tracked down K. Wilber and emailed her, but received no reply.)

After that October post, the angle-bracket-quote becomes more prevalent, but nothing suggests why that message is the ostensible first.

A clue in the comments

Via an Internet history mailing list, Tony Finch reported that he found the closest thing to an explanation in README file that was part of the Unix variant BSD version 4.3, released in 1986. The contributor, Bob Desinger, notes on January 30, 1986:

	I’ve long wanted a way to get someone’s original message into my
	replies, prepended with those nice “> “ netnews-style borders like so:

	> This is the original text that the person wrote.
	> My reply, of course, prepends the original body with the angle-brackets,
	> often called brokets, and skips the headers.

His code change then provides the details of how to make this work. The previous release, 4.2 (1983), didn’t contain this quoted-reply method, which made it seem like this might lead to the origin.

Desinger is still involved with BSD, and he provided useful background on how that change was made, but he believes his inspiration came from Larry Wall, the inventor of perl, but also the creator of rn, the first terminal screen-based newsreader; previous ones were line based. (Wall introduced the original social-media blocking system, too: a KILL file, also known as a bozo filter.)

Wall told me via email, however, that not only can’t he take credit for it, “I recall that I kinda disliked them.” He always used a colon followed by a space, “because I thought it made a prettier line down the screen than angles.” He remembers seeing them before rn, which in any case debuted in 1984, after the early citations I found.

From Wall, I went to Brad Templeton, a Usenet guru and founder of the ahead-of-its-time ClariNet Communications Corporation, which syndicated news articles to Usenet. Templeton noted, as did others, that email at that time — and still in the Unix mbox format — delineates email in a specially reserved way. All lines that start with From mark the separation of one message from the next; mailers insert a > in front of any From that starts a line in the body of a message to avoid breaking this delineation.

Templeton led me to several other folks. Mary Ann Horton, a creator of the B news-reading posting and relay software, consulted the source code of versions 2.9 and 2.10 of around that time. She emailed me, “Neither automatically includes the message being followed up, much less any diples.”

Security guru Steve Bellovin helped create Usenet, but he doesn’t have a recollection. Another key Usenet figure, Gene Spafford, wrote, “I seem to recollect that the behavior came from the mailing lists world first, and then flowed into Usenet, and thence to common usage.”

Bellovin has a notion though: “My assumption is that as reply chains grew longer, the tab that Berkeley Mail was using caused too much indentation, so someone substituted a ‘>’ — but I don’t know who or when.”

The most likely case is that an email program in reasonably wide use between 1981 and 1983 — meaning by a few thousand people — either defaulted to > or allowed it as a quoting option. Because private email isn’t easily accessible outside of perhaps some government archives, and because seemingly no one copied and pasted from such a program into Usenet or other sources, that’s where the trail grows cold for the moment.

Don’t quote me

I’ve been online in some form since 1979, when I used an Ohio Scientific Inc. (OSI) C1P paired with a 110-baud modem to dial up local bulletin board systems. I used Usenet indirectly — it was a store and forward system that percolated posts among nodes — and would have been active during the period that indents switched to quotes, and I have zero recollection as well.

It’s true that not all history can be known, but I feel that there’s a little bit more to this story yet to come.

What song the Syrens sang, or what name Achilles assumed when he hid himself among women, though puzzling questions, are not beyond all conjecture. — Thomas Browne