A post from Shady Characters

The @-symbol, part 2 of 2

This is the most recent in a series of two posts on The @-symbol. Start at PART 1 or view ALL POSTS in the series.


Before its ascent to accidental stardom, the ‘@’ went almost unremarked for centuries. Widely used to mean ‘at the rate of’1 — for example, ‘3 apples @ $1’ is equivalent to ‘3 apples at $1 each’ — the symbol lived out a useful but mundane existence in the world of commerce, rarely warranting a second glance from paleographers or philologists. Even now, when it has been propelled firmly into the limelight by email’s meteoric rise, credible accounts of the symbol’s visual appearance and meaning remain surprisingly thin on the ground.


Emblematic of the lack of pre-email scholarly interest in the ‘at’ sign is its treatment at the hands of Berthold L. Ullman, a prominent 20th century professor of Latin. Ullman’s otherwise comprehensive 1932 treatise on Ancient Writing and Its Influence dismisses the character with a single perfunctory line: “There is also the sign @, which is really for ad, with an exaggerated uncial d.”2 Ad is a Latin preposition meaning ‘to’, or ‘towards’,3 while uncial script was a rounded, uppercase alphabet used in the interregnum between the decline of Roman majuscule and the advent of Caroligian minuscule.4 Unfortunately, inviting though it is to imagine that “AD” might have given rise to ‘@’, Ullman’s confidence in his pronouncement did not extend to providing any evidence for it.

There is a similar theory, and one similarly without documentary proof, that ‘@’ comes from the French à for ‘at’, or ‘at the rate of’, where the scribe would write the letter ‘a’ and then add the accompanying accent without lifting his pen.5 Although this usage is not in question — in at least some French manuscripts, ‘@’ was used in place of à — there is no evidence to suggest that this is the source of the symbol’s shape as opposed to simply another use of the character to mean ‘at the rate of’.


Though the origins of the ‘@’ symbol’s visual appearance are murky at best, its use as a shorthand for ‘at the rate of’ is rather better attested. One scholar in particular saw his work reach a far wider audience than might have been expected of an otherwise minor piece of paleographic research: in 2000, a number of newspapers6,7 reported on the work of one Giorgio Stabile, an Italian academic who had finally unearthed convincing documentary evidence of the symbol’s meaning, if not its visual appearance.

Stabile’s search for the birth of the ‘@’ started with an analysis of the symbol’s various names. A online survey conducted in 1997 revealed that the symbol went by a multitude of names across 37 different countries, many of them playfully inspired by its shape: snabel-a, or ‘(elephant’s) trunk-a’ in Danish and Swedish; apestaart, or ‘monkey’s tail’ in Dutch; zavinác, or ‘rollmop herring’ in Czech and Slovak; Klammeraffe, or ‘spider monkey’ in German; strudel, or a roll-shaped bun, in Hebrew; kukac, or ‘worm’ in Hungarian; grisehale, or ‘pig’s tail’ in Norwegian, and gül, or ‘rose’ in Turkish. French and Italian have both ‘proper’ terms — respectively arobase, an archaic unit of weight, and anfora, or ‘amphora’ — and also the more whimsical escargot and chiocciola, both meaning ‘snail’. English uses the cheerlessly direct ‘commercial at’ or, simply, the ‘at sign’.8

Stabile observed that despite the symbol’s many metaphorical aliases, only certain names stood out as unrelated to its shape: the English ‘commercial at’, the French arobase (also rendered in Spanish and Portuguese as arroba), and the Italian anfora, or ‘amphora’. ‘Commercial at’ evidently described the character’s typical usage, but arobase/arroba and ‘amphora’ bore further investigation.9

Amphorae were long-necked pottery storage jars with tapered bases, used for centuries by the Greeks and Romans to transport cereals, olives, oil and wine,10 and the word ‘amphora’ referred not only to the vessels themselves but also related units of volume and weight. The standard Roman amphora, embodied in the ‘amphora Capitolina’ kept securely in Rome itself,11 had a volume of a cubic foot, or approximately 26 litres. The Spanish and Portuguese arroba, on the other hand, was a customary unit of weight and volume, representing either a quarter of a quintal,12 or hundredweight, or alternatively a volume of around 16 litres of liquid.* The word itself came from the Arabic al rubʽ, or ‘one fourth’,13 a term absorbed into the languages of the Iberian peninsula during their period under Moorish rule, and which later made its into French as arobase.

The key to Stabile’s discovery was a letter sent from Seville to Rome on May 4th 1536 by a merchant named Francesco Lapi, in which Lapi discussed the arrival in Spain of three trading ships from the New World. Writing that an amphora of wine sold there for 70 or 80 ducats, he used the familiar ‘@’ symbol as an abbreviation for that word:9

'@' for 'amphora' in Francesco Lapi's letter of May 4th, 1536
‘@’ for amphora in Francesco Lapi’s letter of May 4th, 1536. (Image taken from Documenti per la storia economica (Secc. XIII-XVI) by Federigo Melis, via The @ Book by Patrik Sneyd. Permission to reproduce granted by Firenze University Press.)

Consulting a contemporaneous Spanish-Latin dictionary,14 Stabile found that arroba was synonymous with amphora; the Spanish, Portuguese and Latin units of measure might have differed in their exact definition, but the ‘@’ was their common shorthand. In those southern European countries, then, the ‘@’ symbol is named for and embodies a link to its ancient roots: arroba, anfora and arobase are perhaps the character’s truest names.


From its starting point as a unit of measure in southern Europe, the ‘@’ became shorthand for any unit in the northern part of the continent, giving rise to its English language name of ‘commercial at’ and its meaning of ‘at the rate of’.9 Prospering in commercial circles as a handwritten and printed character,16 the 19th century saw the arrival of two innovations which together would change the face of writing, printing and business, and the ‘@’ symbol would have to move with the times in order to survive.

The first came in 1868, when an American inventor named Christopher Latham Sholes received two patents relating to “that class of machines designed to write with types instead of a pen”:17 though not the first to try,18 Sholes had succeeded in developing what would become the first commercially successful typewriter.19 This new machine, with which anyone could quickly produce regular, legible documents, revolutionised clerical work and would become the default method of text entry for more than a hundred years.

The ‘@’ symbol, however, did not make it onto the keyboard of Sholes’ prototype typewriter, nor that of the first commercial variant produced by Remington a few years later. The piano-like keyboard of the 1867 prototype bore two rows of keys, with uppercase letters arrayed along the bottom row and the numbers 2-9 (the letters ‘O’ and ‘I’ stood in for zero and one) and a scant few miscellaneous glyphs (; $ – . , ? /) along the upper row.20 The later Remington model bore a four-row keyboard with an arrangement close to the modern QWERTY layout, but still the ‘@’ symbol was absent.21 The inclusion of the dollar sign indicates that Sholes was aware of his machine’s commercial applications, but despite its practical utility the ‘@’ would have to wait a few more years until it was picked up by other manufacturers, such as on this 1889 Hammond 1:

The '@' symbol on the keyboard of an 1889 Hammond typewriter
The ‘@’ symbol shares a key with ‘X’ and ‘⅞’ on the keyboard of an 1889 Hammond typewriter. (Images courtesy of Richard Polt.)

Once adopted as part of the increasingly standardised typewriter keyboard, the ‘@’ symbol’s immediate future was assured. However, the next innovation to shake the world of information processing, such as it was in the late 19th century, came only a few years later. The results of the United States’ 1890 census were collated using a baroque new electro-mechanical device called the Tabulator, designed and built by a statistician named Herman Hollerith.22 Like Sholes, Hollerith was not the first in his field but he was the most successful thus far: his Tabulator foreshadowed the programmable computer, and his use of punched cards to provide it with data would persist until the 1970s.

Hollerith’s cards held twenty-four columns of twelve rows each and were specifically designed to record census information for one individual.22 When used for general data entry instead, each column represented a digit in the range 0-9 and two optional ‘control’ positions, used to indicate special conditions such as a credit balance.23 With only numeric characters supported, the ‘@’ did not get a look in, and despite its status as standard character for text input, the symbol would again have to fight its way into new standards of text storage. A typist could enter an ‘@’ on their keyboard, but to a computer which did not understand it, the ‘at’ sign did not even exist.

An original Hollerith punched card
An original Hollerith punched card, circa 1895, printed to highlight the census-specific information carried by each position. (Out of copyright image taken from Wikimedia Commons.)

Little by little, the repertoire of characters which could be represented on a punched card grew larger. By 1932, for instance, a ‘Hollerith card’ punched according to the so-called Binary Coded Decimal Interchange Code (BCDIC) could draw from a set of forty characters: the digits 0-9, the letters A-Z, the minus sign, the asterisk, the ampersand and a space.23 BCDIC grew again in the 1950s, now encompassing forty-eight characters in total, and with this expansion the ‘@’ finally took its place alongside other non-alphabetic symbols such as the octothorpe, dollar sign and percent sign.23 The ‘@’ began to appear in other coding schemes too. Though the United States Army’s 1960s FIELDATA code did not include it,24 its civilian counterpart gave it pride of place as the first character in the code,25 while IBM (descended from Herman Hollerith’s original Tabulating Machine Company26) used the character in its influential 1961 ‘Stretch’ supercomputer.27

Understandably, this proliferation of coding schemes caused some consternation in computing circles, and led to the creation in 1963 of the American Standard Code for Information Interchange (ASCII), a standardised character set intended for worldwide use.28 By now a regular in many other coding schemes, the ‘@’ was an obvious candidate for inclusion. Thus, when Ray Tomlinson looked down at his teletype in 1971 and mulled over the symbols available to him, the combination of a standard keyboard and a standard character set made it inevitable that the ‘@’ key would be there to meet his gaze.§


As email (and the Internet as a whole) grew more prominent, ‘@’ came to symbolise not only the Internet itself but also a sort of generic modernity, or progress: at the turn of the 21st century, the symbol could be found in the names of everything from internet service providers (Excite@Home) and internet ‘c@fés’ to plays (‘F@ust, Version 3.0’30).

Now, though, some years after the dot-com bubble knocked the wind out of the first wave of internet entrepreneurs, the ‘@’ has once again become commonplace, and an ever-expanding roster of replacements are taking over its mantle. No longer is there a universal symbol of connectedness or modernity, and this survivor from the first days of the internet is giving way to unthreateningly generic ‘e-’ and ‘i-’ prefixes which owe more to marketing departments than technical innovation. The ‘@’ has once again become common currency, although it can rightly claim a scope rather greater than purchase orders and grocers’ chalkboards.

1.
Bringhurst, Robert. “At”. In The Elements of Typographic Style : Version 3.2, 303+. Hartley and Marks, Publishers, 2008.

 

2.
Ullman, B. L. “Abbreviations and Ligatures”. In Ancient Writing and Its Influence. Cooper Square Publishers, 1963.

 

3.
Leverett, F. P., Jacobo Facciolati, Egidio Forcellini, Imm Scheller, G. H. Lünemann, and H. W. Torrey. A New and Copious Lexicon of the Latin Language. J.H. Wilkins and R.B. Carter; C.C. Little and James Brown, 1837.

 

4.
Bischoff, Bernhard, and . “Uncial”. In Latin Palaeography: Antiquity and the Middle Ages, 66-72. Cambridge University Press, 1995.

 

5.
Long, Tony. “May 4, 1536: C U @ the Piazza | This Day In Tech”. Condé Nast Digital, May 2009.

 

6.
Willan, Phillip. “merchant@florence Wrote It First 500 Years ago”. The Guardian. July 1, 2000.

 

7.
Olivero, Dario. “Quella Chiocciola Antica Inventata Dagli Italiani”. Gruppo Editoriale L’Espresso, July 28, 2000.

 

8.
Chung, Karen S. “Summary: The @ Symbol”. The Linguist List 7, no. 968 (July 2, 1996).

 

9.
Stabile, Giorgio. “L’icon@ Dei Mercanti”. Istituto della Enciclopedia Italiana, July 31, 2011.

 

10.
Encyclopaedia Britannica. “Amphora (pottery)”.

 

11.
Smith, William. “Quadrantal”. In Dictionary of Greek and Roman Antiquities, 979-980. London: Walton and Maberly, 1853.

 

12.
“Quintal”. In Merriam-Webster Online. Springfield, Massachusetts: Merriam-Webster, Incorporated, July 2011.

 

13.
“Arroba”. In Merriam-Webster Online. Springfield, Massachusetts: Merriam-Webster, Incorporated, July 2011.

 

14.
Gaya, S G. Tesoro Lexicografico, 1492-1726. Consejo Superior de Investigaciones Científicas, Patronato "Menéndez Pelayo," "Antonio de Nebrija,", 1947.

 

15.

 

16.
Pasko, W W. “A”. In American Dictionary of Printing and Bookmaking, 1+. H. Lockwood, 1894.

 

17.
Sholes, C Latham, Carlos Glidden, and Samuel W Soulé. “Improvement in Type-Writing Machines. U.S. Patent 79,265”. US Patent and Trademark Office, July 14, 1868.

 

18.
Pratt, John. “Type Writing Machine”. Scientific American 1017, no. 1 (July 6, 1867).

 

19.
Day, L, and I McNeil. “Sholes, Christopher Latham”. In Biographical Dictionary of the History of Technology, 1104-1105. Taylor and Francis, 2003.

 

20.
Weller, C. E. “Home of First Typewriter”. In The Early History of the Typewriter, 20-21.

 

21.
Bliven, B. The Wonderful Writing Machine. Random House, 1954.

 

22.
Heide, L. Punched-Card Systems and the Early Information Explosion, 1880–1945. Johns Hopkins University Press, 2009.

 

23.
MacKenzie, Charles E. “Early Codes : BCDIC”. In Coded Character Sets: History and Development, 66-67. Addison-Wesley Pub (Sd), 1980.

 

24.
MacKenzie, Charles E. “Early Codes : FIELDATA”. In Coded Character Sets: History and Development, 64-66. Addison-Wesley Pub (Sd), 1980.

 

25.
“Univac Fieldata Codes”. National Aeronautics and Space Administration, August 7, 2011.

 

26.
Aul, William R. “Herman Hollerith: Data Processing Pioneer”. Think, November 1974, 22-24.

 

27.
MacKenzie, Charles E. “Early Codes : The Stretch Code”. In Coded Character Sets: History and Development, 67-75. Addison-Wesley Pub (Sd), 1980.

 

28.

 

29.
Cochran, Shannon. “Morse Code Meets the Internet”. Dr. Dobb’s Journal: Software Tools for the Professional Programmer 29, no. v. 29 (May 1, 2004).

 

30.

 

*
The Spanish and Portuguese units of weight are actually slightly different: the Spanish arroba was around 25 pounds, while the Portuguese unit was 32 pounds.13 
Giorgio Stabile’s discovery is corroborated by Jorge Romance, who in 2009 uncovered an even earlier use of the ‘@’ symbol in a Castilian document from 1445.15 
Sholes’ invention was so successful and so pervasive that the interrobang’s appearance on a Remington typewriter keyboard was still considered newsworthy almost a century later. 
§
Ironically, the ‘@’ symbol was only inducted into Morse code — the grandaddy of encoding schemes — as recently as 2004.29 

24 comments on “The @-symbol, part 2 of 2

  1. Comment posted by The Modesto Kid on

    You mention that the earliest typewriters had no zero or one key, and I was recently reading this article in which modern-day college students using a typewriter are surprised by the lack of a one key. Do you know when zero and one became standard on typewriter keyboards? I learned to type on an electric typewriter which must have been from my parents’ college days or shortly thereafter, so circa 1970, and it certainly had both numerals. Is this a question of electric vs. manual typewriters? I never have used a manual typewriter, that I can recall.

    1. Comment posted by Keith Houston on

      Hi MK — Richard Polt has commented below on that very topic. You might want to take a look Richard’s website too — it’s a great source of information on vintage typewriters.

  2. Comment posted by Schesis on

    Dunno about Spain, but the arroba is still used as a measure here in Bolivia – and it’s still about 12kg / 25lb when used as a measure of weight. It’s kinda vague, though – it more or less means “a sackful” here, so an arroba of potatoes, for example, can vary a fair bit.

    1. Comment posted by Keith Houston on

      Hi Schesis,

      I did get the feeling that the arroba was and is a very South American unit. Is the ‘@’ symbol still used to mean arroba in the sense of the unit of weight, or is it confined to ‘at the rate of’?

    2. Comment posted by Aaron Davies on

      Sounds a bit like league/legua, which apparently in much of Latin America still literally means “about an hour’s walk (accounting for local terrain)”.

    3. Comment posted by Keith Houston on

      Hi Aaron — there does seem to be a lot of sense in using “human”-sized measurements, quantities that are instinctively understood because they relate so well to our own experience. As I software engineer I naturally gravitate towards the metric system, but I still find it easier to estimate feet and pounds than metres and kilograms.

      Thanks for the comment!

  3. Comment posted by Richard Polt on

    The availability of zero and one on manual typewriter keyboards varies from model to model. In general it’s a more luxurious feature, since strictly it is not necessary. Thus larger and more expensive typewriters are more likely to have them. I think zero is more common than one.

    1. Comment posted by The Modesto Kid on

      Good to know — thanks for the info, Mr. Polt. I linked your site from the sidebar of READIN.

  4. Comment posted by Rainer Brockerhoff on

    The old typewriter I learned to type on also has only the 2 to 9 numerals, uppercase ‘O’ was used as ‘zero’, but the stand-in for ‘one’ was ‘l’ (lowercase ‘L’). Indeed, if you look at a font like American Typewriter, 1 and l look identical for most practical purposes.

  5. Comment posted by Diane on

    I turn with relief to your blog. Like the nicest people it has clarity, directness and yet conveys a sense of the unspoken. I leave feeling more appreciative – not only of this converse but, inexplicably, of the world in general.

    1. Comment posted by Keith Houston on

      Hi Diane — I’m glad you’re enjoying the site. I’m still amazed at (and grateful for!) all the positive responses to it.

  6. Comment posted by Dr. Mabuse on

    I always thought that “@” was pronounced “circa”, and that it probably derived from shorthand – the first and last letters combined into one symbol. Anyway, that’s the way I’ve always used it, and it’s pretty close to “at”.

    1. Comment posted by Robert Allwood on

      Sounds more like the copyright sign. Circa (about) is usually abbreviated as ‘c.’.

  7. Comment posted by Richard Jenkins on

    Continuing with the folk etymology, I was taught “@” on grocers’ signs read as “each” (a simpler way of saying of “at the rate of”). One could view the @ character as an “e” containing smaller “a,” corresponding to the abbreviation “ea.”

    1. Comment posted by Keith Houston on

      Hi Richard,

      It’s interesting you should bring up ‘ea.’ in connection with the ‘@’ symbol. Richard Polt, who has been helpful in answering my questions about typewriters and who provided the images of the Hammond 1 above, mentioned in an email that he has seen at least one ’60s typewriter with an ‘ea.’ key. I haven’t come across any suggestions of an etymological connection, but it’s quite a coincidence.

      Thanks for the comment!

  8. Comment posted by Alexander on

    The amphora theory is a nice one. But then what about the usage of “@” as the initial for the first letter of “amen”, such as given in this image (http://en.wikipedia.org/wiki/File:19-manasses-chronicle.jpg) accompanying the corresponding Wikipedia article (http://en.wikipedia.org/wiki/At_sign)?

    One can see the “@” there very clearly, in its modern form. So obviously it was in use before the 16th century, and not only for amphoras. Or was it just that particular scribe just this once?

  9. Comment posted by Michael Last on

    Some 15 years ago I was told that the original use of the @-symbol was as a religious symbol for “Anno Domini” – the “a” being inside a “D”.

    I have seen the @-symbol engraved in a very old baptismal font in a church in Copenhagen, Denmark. The symbol I saw was looking very much like the @-symbol we use today as the email-symbol.
    (“Anno Domini” meaning “in the year of our Lord” should be the reason it was/is to find on some baptismal fonts.)

    Unfortunately I cannot find any information online to back up this theory.
    Have anyone ever heard/read about this “@-symbol theory” before ?

    1. Comment posted by Keith Houston on

      Hi Michael — that’s an intriguing theory. I can’t say I’ve come across it before. Do you happen to remember when the church was built, or how old the font was?

    2. Comment posted by Michael Last on

      Hi Keith – I only remember that it was an old church, could be maybe 200 years old or so, also the font. But as mentioned I’m not able to find any information about it anywhere and since I’m not living in Denmark anymore, I can’t start walking from church to church checking it out : )

    3. Comment posted by Keith Houston on

      Hi Michael — ah, I see. Thanks all the same! The origins of the ‘@’ seem to become more and more confused as time goes by.

  10. Comment posted by Anneke Myers on

    Hi Keith,
    I’ve just tonight come across your excellent series on the @ symbol, thank you! Very much enjoyed it.

    Also, I have a thought on the 14th century ‘Amen’ example referred to in the comments. Perhaps this is in fact a ligature of the lower case alpha and omega characters, referencing Jesus (referred to in the Bible as ‘the Alpha and the Omega’).

    1. Comment posted by Keith Houston on

      Hi Anneke — thanks for the comment! I’m glad you enjoyed the series.

      I just took another look at the “amen” example you mentioned, and it certainly looks like it could be a ligature of the alpha and omega characters, but, like so many other potential examples, I suspect we’d need a real live palaeographer to tell us for sure.

      Thanks again!

Leave a Reply to Keith Houston Cancel reply

Required fields are marked *. Your email address will not be published. If you prefer to contact me privately, please see the Contact page.

Leave a blank line for a new paragraph. You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>. Learn how your com­ment data is pro­cessed.