Yesterday, in discussing Kevin Fowler's song Pound Sign, there was some debate about the origin of the term "pound sign" for the symbol #. I suggested that it all started with the substitution of # for £ on American typewriter keyboards, but others argued that # was a standard symbol for pound(s) avoirdupois. I've heard this theory before, but I expressed skepticism about it because I've never actually seen the symbol used that way.
Today, after some further research, I'm still not completely sure. But I've found a new theory, which I think has a better chance to be correct: it's all Emile Baudot's fault.
William Safire asked about this usage in his On Language column for 1/20/1991, "MAIL CALL (ALL CAPS)":
… you and I rarely use suite or even apartment number , abbreviated to Apt. No. ; most often, we use the crosshatch symbol for "number," which has come to be known as the pound sign (the origin and etymology of which I seek Lex Irreg help in finding).
Others writing about the same time indicate that this usage was unfamiliar to them as well. Thus Ron Alexander, "Metropolitan Diary", NYT 6/20/1990:
The responses are delivered in a reassuring, synthesized male voice that never hesitates or sneezes or coughs. It says to "push the pound sign after entering the deposit amount." The pound sign, the voice explains, "is to the right of the zero; the asterisk is to the left of the zero."
I believe that AT&T used "pound sign" for this symbol from the first introduction of the touch-tone keypad in 1963, but it's clear that in 1990-1991 the term was still new to many Americans. I left AT&T in 1990, and at that point, "touch-tone penetration" (as we used to call it) was still low enough to be an argument for outfitting voice response systems with speech recognition technology, even though that technology was then very expensive and not very robust.
But touch-tone penetration (the proportion of households with touch-tone as opposed to rotary phones) started to pick up after 1984, when customers could buy their own phones rather than leasing them from AT&T as part of their service package. And by 1990 or so, voice response systems were starting to take off, and people throughout the country were forced to learn what a "pound sign" was.
Safire reported the results of his query a couple of months laters, in "Hit the Pound Sign", 3/24/1991. He starts with a fictionalized voice-mail interaction:
"Please enter your password," the synthetic-syrupy recorded voice directs, "followed by the pound sign."
"When was I born?" is my first internal question. The second, which I asked only once, was: "What and where is the pound sign ?"
I always thought the pound sign was the symbol for the British pound sterling, a script capital L with a hyphen through it, a stylized representation of the Latin libra . (The Roman pound was several ounces lighter than the modern pound; some things stick in my mind, but numbers fade fast.)
No; the sign that the recorded Miss Syrup refers to is what some of us remember as the tick-tack-toe sign, or the crosshatch, or the sign of the double-cross, similar to the symbol Charlie Chaplin wore as a parody of the swastika in his 1940 movie, "The Great Dictator."
After some more stuff making fun of voice menus, Safire gives some of the things that are not answers:
When #, as we shall call it here, is placed before a number, it is called a number sign ; #1 pencils have a softer lead than #2 pencils, and nothing fits into a #10 envelope anymore. When a musician sees it, the meaning is "sharp"; doctors often use it as a symbol for "fracture."
When a proofreader uses the #, the meaning is "insert space here": two words incorrectly runtogether are happily separated by a #.
And finally he gives the answer that he's settled on:
The origin [of the touch-tone keypad term] may be from the use of # to mean "pound," as in "a 5# bag of sugar," written by someone unhappy with the abbreviation lb. to stand for "pound." (We know that pound is from the Latin libra pondo , "a pound by weight," which accounts for the lb. ; not everybody knows that) [...]
A more remote possibility is that pound evokes a mashing of the desired button, as one pounds on a door, but I go for the 5#-bag theory.
The trouble with this theory, as I observed in my earlier post, is that the evidence for # meaning "pound avoirdupois" is so thin. It's clear that Safire himself had never encountered it, or he wouldn't have needed to ask for help from his readers. I myself have never seen it in a grocery store or a recipe or a newspaper ad — and I've been shopping and cooking for myself since 1963.
Lists of abbreviations from the late 19th and early 20th century don't use it:
Thus N.A. Calkins, Manual of object-teaching, New York, Harper & Brothers, 1882:
Or Bernard Fantus, A text book on prescription-writing and pharmacy, 1905 ("Second Edition, Thoroughly Revised and Adapted to the Eighth (1905) Edition of the the United States Pharmacopoeia"):
That last reference is especially relevant because it uses non-standard typography for several other measures, but gives "pound" as simple "lb."
The earliest example that I've been able to find of the symbol # apparently used as shorthand for "pound (as a measure of weight)" is the OED's citation:
1923 W. E. HARNED Typewriting Stud. II. 29/1 Special Signs and Characters..#..Number or pound sign; # 10 (No. 10); 10# (ten pounds).
A similar reference can be found in Nancy Lawrence et al., Correlated studies in stenography:
the correlation of business correspondence, English, office practice, and shorthand, 1932:
But these might literally refer to some rather specialized kind of business shorthand, rather than a general usage. There's a more helpful example in Milton Kaufman, Radio operator's license Q & A Manual, 1955:
Similarly, the number or pound sign (£) cannot be reproduced in transmission and in lieu thereof the sender should use the words NUMBER (or NO), POUND (or LB) or POUNDS (or LBS), as the case may be.
Baudot invented his original code during 1870 and patented it during 1874. It was a 5-bit code, with equal on and off intervals, which allowed telegraph transmission of the Roman alphabet and punctuation and control signals. It was based on an earlier code developed by Carl Friedrich Gauss and Wilhelm Weber in 1834.
The code was entered on a keyboard which had just five piano type keys, operated with two fingers of the left hand and three fingers of the right hand. Once the keys had been pressed they were locked down until mechanical contacts in a distributor unit passed over the sector connected to that particular keyboard, when the keyboard was unlocked ready for the next character to be entered, with an audible click (known as the "cadence signal") to warn the operator. Operators had to maintain a steady rhythm, and the usual speed of operation was 30 words per minute. [...]
Since it was a five-bit code, 32 symbols could be represented. In order to extend this set, there were two mode symbols, allowing the system to shift into "letters mode" or "figures mode".
(The following details are taken from this page, perhaps with some scribal or interpretive errors on my part.)
In the version of the code adopted as ITA 1, several of the "figures mode" symbols were left "at the disposal of each administration for its internal service". One of these was 11011, which in the version of Baudot used by the British Post Office was N in letters mode, and £ in figures mode. I believe that in U.S. versions of ITA 1, this was # (though I haven't been able to find a reference to nail this down).
In the version created by Donald Murray in 1901, the five-bit pattern 10111 was the letter X in "letters" mode, but was the pound sign £ in figures mode (that is, after a figures shift sign 11011 had been sent). In America, the Teletype Corporation adopted Murray's system with some changes, one of which was to substitute # for £ as the meaning of 10111 in figures mode.
In the ITA 2 five-bit code (in use since 1930 or so), the details are again different, but the relevant correspondences are the same. In letters mode, 10100 is the letter H, but after a figures shift shift 11011 has been sent, it's £ in international versions of the code, and # in American versions — called "pound sign" in both cases.
So I'm quite sure that this is why the engineers at Bell Labs called # "pound sign" — it corresponded to a Baudot code-point that had been used for £ in the UK and # in the U.S., probably since the late 19th century and certainly since the early 20th century. Those guys would have memorized the ITA 2 tables, without any question.
This is probably also why typewriter keyboards shared this meaning for shift-3.
A few questions remain.
1. When was the Baudot figures-mode correspondence between £ and # first established?
2. When the correspondence was first established, was it essentially random? That is, did some Americans say to themselves, "We don't need that £ symbol, so what should we use that code-point for? How about §? Naah. Well then, what about #? Sure, why not." Or instead, was there already an established usage of # for pound(s) avoirdupois, so that it was a natural pun?
My money is on the "random code-point re-use" theory. And on that view, any sporadic use of # to mean "pound(s) avoirdupois" would be a secondary and derivative effect, not a cause.
[By the way, I once spent a week in Denver recording the young woman that Bill Safire called "Miss Syrup". But that's a story for another time.]