Thousands of Years Ago.

Trying to finish last week’s New Yorker before the new one arrives, I began “The End of Ice: Exploring a Himalayan glacier” by the estimable Dexter Filkins and was pulled up short before I finished the first paragraph:

The journey to the Chhota Shigri Glacier, in the Himalayan peaks of northern India, begins thousands of feet below, in New Delhi—a city of twenty-five million people, where smoke from diesel trucks and cow-dung fires dims the sky and where the temperature on a hot summer day can reach a hundred and fifteen degrees. The route passes through a churning sprawl of low-land cities, home to some fifty million people, until the Himalayas come into view: a steep wall rising above the plains, the product of a tectonic collision that began thousands of years ago and is still under way.

“Thousands of years ago?” I thought. “What the hell?” I looked it up, and found what I expected: “The Himalayan mountain range and Tibetan plateau have formed as a result of the collision between the Indian Plate and Eurasian Plate which began 50 million years ago and continues today.” But then it occurred to me that, technically, “thousands of years ago” is perfectly correct — it’s just a whole lot of thousands. And I reflected on how it is we use such words; there’s no fixed amount at which it stops making sense to say “thousands” (you could certainly use it of something that happened, say, 12,000 years ago, but not a million years ago, and I personally would stop well before the 100,000-year mark), but that doesn’t mean it’s endlessly flexible. I guess it’s a sorites problem.

At any rate, when I went to the online version of the article to copy the paragraph, I found it had been rewritten to say “a tectonic collision that began millions* of years ago,” with a footnote reading: “An earlier version of this article incorrectly identified when the tectonic collision that formed the Himalayas began.” So for the New Yorker, at any rate, “thousands” here wasn’t just absurdly understated, it was out-and-out incorrect.


  1. Yeah, there are a lot of cases where an “expansive” reading is notionally possible, but precluded by contextual or pragmatic factors. There’s the famous Mitch Hedberg joke, “I used to do drugs. I still do, but I used to, too.”

    It’s also occurred to me that the widespread injunction “Drink responsibly” could be read as commanding everyone to drink – albeit responsibly. But it’s always intended with an implied conditional: “If you drink, do so responsibly”.

  2. I’ll take “scalar implicature” for $1,000, Alex.

  3. I recall an activist being interviewed on NPR in the early 1990s. She said that some humanitarian crisis I had been only vaguely aware of was possibly the worst in the world at the time, with “scores” of deaths. I remarked to my father that “scores” suggested less than one hundred and should certainly mean less than two hundred. That made the deaths, while undoubtedly tragic, minuscule next to what was happening in the former Yugoslavia (among other locations).

  4. I’ll take “scalar implicature” for $1,000, Alex.

    Thanks, I knew there had to be a term for it!

  5. “Score” probably does not mean 20 anymore. Just a “large number”.

  6. True, although most Americans should at least be familiar with the older meaning if they learned about the Gettysburg Address in school.

  7. Or if they foresee running out of weed in an unfamiliar city.

  8. Squiffy-Marie von Bladet says

    I think it was DH Lawrence who famously said:

    Some time before last Tuesday,
    the Roman Empire fell.
    Exactly how much time before,
    I really couldn’t tell.

    That’s the empire in the West, at least,
    I’ve heard some people say
    that Byzantium, in the East
    held out till the other day.

  9. I have to admit being brought up short while reading James S.A. Corey’s Leviathan Wakes when an asteroid was described as following “its millennia-long journey around the sun”.

    Well, yes, something that’s been going on for four and half billion years has been going on for “millennia”, but still…

  10. SFReader says

    there is a suitable non-word for that – billenia…

  11. Rodger C says

    Geocenturies. (Periods of 100,000,000 years.)

  12. SFReader says

    In astronomy an aeon is defined as a billion years (10^9 years, abbreviated AE).

    Example sentences

    This mass of molten continent still, however, retains enough homogeneity to be returned more or less as a unit at the mid-ocean ridges eons later as the rock cycle continues.

    Astronomers believe that comets and asteroids hitting the Moon eons ago left some water behind.

  13. SFReader says

    Aeon appears an American spelling, but I’ve never seen the word spelled other than eon.

    Because I read too much British SF, I guess.

  14. SFReader, I don’t think your reading is a representative sample. Looking at Google Ngrams, I see the spelling preference going the other way, as one would expect from other pairs in which British English is more likely to have “ae” where American English has “e”. H.P. Lovecraft isn’t typical of US English, especially modern US English.

  15. Even if if it has a “formal” definition, “aeon” isn’t actually a word used by astronomers in serious discussions. Like “light year,” it is only used in discussions aimed at people outside the field. (Real astronomers always deal in parsecs.) Terms like this abound in science, actually. They were introduced by genuine scientists in an attempt to provide useful technical vocabulary. However, they did not survive in the language scientists actually use. Yet they persist in popular accounts, perhaps because they seem to be simpler in some way than the real technical terminology of the field.

    As to the original asteroid, I would interpret “millennia-long journey around the sun” as a reference to its orbital period, not the total time the asteroid had been in ballistic motion.

  16. @Brett:

    You’re quite right about astronomers not using “aeon/eon”. “Gigayear” (Gyr) is moderately common.

    As for the orbital period: the problem is that the asteroid in question was part of the main belt, orbiting between Mars and Jupiter. The orbital periods of those two planets are 1.9 and 12 years, respectively, which sets the limits on the possible orbital period of the asteroid (most have periods of 3-6 years).

    So… micro-millennia, yes, regular millennia, no.

  17. Oh, wait, I guess “micro-millennia” would be days; I should have said “milli-millennia”…

  18. There are about π seconds in a nanocentury.

  19. J. W. Brewer says

    One can think of instances where one might deliberately use an out-of-scale formulation for rhetorical effect, e.g. saying something happened “hundreds of months” or “thousands of hours” ago, which make the oddity/incongruity explicit. (I might myself say for rhetorical effect that it’s now been “sixty semesters” since I studied Homeric Greek and ergativity in Dyirbal . . .) But I guess this wasn’t an instance of that.

  20. Dozens of people were hurt in the Battle of the Somme.

  21. Larry Niven uses the oddball scalar expression “half a thousand years” several times:

    In the continuous roar there was something hypnotic, something that might have explained why the crowded-room drinking bout was more than half a thousand years old; for monotonous background noise has long been used in hypnosis.

    For more than half a thousand years those laws had remained the same.

    Louis knew that symbol, a trademark half a thousand years old.

    Boosterspice had kept some humans hale and sapient for half a thousand years, sometimes more.

    Some Kzinti can perceive hyperspace directly; their female kin have mated into the family of the Patriarch for half a thousand years.

    What they had seen, half a thousand years ago, was a perfect smoke ring several tens of thousands of kilometers across, with a tiny hot pinpoint in its center.

    Human crew were supposed to be in that cycle, and they had been gone for half a thousand years.

    “For the last half a thousand years, the only thing that’s kept Atlantis above the waves has been the spells of the sorcerer-kings.”

    “He hasn’t been heard of in half a thousand years, not since the Nordiks [sic] conquered the Frost Giants.”

    Henceforth the capital was secure against submersion, but another danger, permanent and much more subtle, remained, viz., the unhealthiness of the ground itself, sodden as it was with the filth of half a thousand years.

    (Okay, I lied; the last one is by Hans Gadow, and popped up because of a hit on a different Niven altogether.)

  22. I can see using it once for effect, but it’s an odd expression to have as a tic! Reminds me of the Dutch novel I just read, In Babylon, where I swear every single character at one point or another raised or lowered one eyebrow.

  23. I’m a fan of the Song of Ice and Fire series, but good Lord does George R. R. Martin love his pseudo-archaisms and stock expressions. One character – a woman grown, clad in boiled leather and a coat of rusted mail – spends what feels like half a hundred years (or near enough as makes no matter) trying to locate a maid of three-and-ten, but her efforts prove as useless as nipples on a breastplate – little more than a cruel jape or a mummer’s farce.

  24. Augustus Maria von und zu Blattburg says

    It was definitely Stalin who said “to lose one parent is a tragedy, to lose several is a statistical anomaly”.

  25. raised or lowered one eyebrow


  26. David Marjanović says

    where I swear every single character at one point or another raised or lowered one eyebrow

    I wonder if that many people can even do this; I can’t – it’s both or neither for me.

    (On the other hand, I have no trouble whatsoever with the Vulcan salute. :-þ )

  27. George Gibbard says

    I can wiggle my left ear up and down, but not my right.

    My grandfather’s ears wiggled when he chewed. The same is true of pandas.

  28. There are some odd in-group expressions that kind of make sense if you are in the know. I’ve heard, economists measure certain things every quarter (that is, 3 months, 1/4 of a year) and may discuss what happened, for example, for the last 25 quarters. Which, I guess, would be quite insane to say in any other context.

  29. My impression, which could be wrong, is that UK speakers are more likely than Americans to talk about, say, 4,500 as “four and a half thousand” rather than “four thousand five hundred” or “forty-five hundred”, but I think that’s a separate issue from calling 500 “half a thousand”.

  30. @John Cowan: I’m not sure if this was intentional, but saying that Niven used a phrase “several times” and then giving ten examples seems itself (to my ear) an instance of this phenomenon in question.

  31. Originally I only had #2-#5, and then it occurred to me to widen my search, but I forgot to change several.

  32. Whereas 4500 might be “forty-five hundred”, 4,500 must be “four thousand five hundred”.

  33. SFReader says

    “two hundred and twenty five scores” or “three hundred and seventy five dozen” or “forty five fivescores”

  34. Whereas 4500 might be “forty-five hundred”, 4,500 must be “four thousand five hundred”.

    I don’t make that distinction; I’d be equally likely to read the latter either way.

  35. My previous comment ought to have made clear that it was only my personal interpretation. Having said which, in Ireland I only hear “n hundred” for n > 9 in “fifteen hundred metres” as the name of a race (running or swimming).

    I associate the “n hundred hours” time convention with the *US* military, but my military experience is limited to movies, mainly American.

    I do approve of the convention of omitting the thousands-separator comma from four-digit numbers.

  36. This seems like a good place to post that I’ve finally stabilized my convention for pronouncing 21C year numbers: 2000 is “two thousand”, 2001-09 are “two thousand and one” to “two thousand and nine”, and 2010-2099 are “twenty ten” to “twenty ninety-nine”. I still don’t have a name for the 2000-2009 decade as a whole, though.

    Nevertheless, I still say “nineteen oh one” to “nineteen oh nine” for 1901-1909, but there is something about “twenty oh” that seems Wrong. Abstractly, I like the convention of calling 1801 “the year one” and so on, but nobody’s done that for two centuries now, and it would just be confusing.

    (Yes, weenies, I know 2000 is part of the 20C. I don’t care. The Italians have it right anyway: 1300-1399 is the Trecento, and so on.)

  37. J. W. Brewer says

    I think the whole humor (traditional at least among math/science/engineering types?) of “furlongs per fortnight” as a jocular measure of velocity depends on the same pragmatic intuition as to the incongruity of the choice of scale (more particularly that the two halves don’t seem to belong in the same scalar context), and I expect there are other examples.

  38. J. W. Brewer says

    But one can also find multiple examples where technical usage is at variance with popular. E.g. the nurse may write down your height on a medical form as “71 inches” even though in a lay context that would be a very weird way to say you were “five foot eleven.” And in military circles I believe it was and perhaps still is (if they haven’t metricated) usual to quantify artillery range in yards, even when the numbers get high, e.g. you might say that the main guns on such and such battleship could reach a target from 16,000 yards (not [approx] nine miles) away.

  39. “noughties” has been standard in Britain and Ireland since the late noughties, and appears in parliamentary transcripts without apologetic quotation marks.

    Our names for post-2000 years were settled around the same time and match John Cowan’s. Are there many USians who contrast “two thousand six” [for the number] with “two thousand *and* six” [for the year]?

  40. I don’t know. Omitting “and” in the names of integers is prescriptive AmE (implanted by math teachers rather than English teachers), and I am too old, or perhaps too bullheaded, to accept this change from the natural English usage of “five thousand two hundred and eighty” for the number of feet in a mile.

  41. SFReader–
    “two hundred and twenty five scores” sounds to me like scores on a scoreboard [= units].
    Lincoln didn’t say “four scores and…”
    “three score and ten”
    Surely “score” doesn’t take an s to become plural, does it?

  42. @John Cowan: Yeugh, I hate that one. My batty fourth-grade teacher told us not to say and because it represents a decimal point – as if anyone on Earth would say “one and three” in place of “one point three”. You can take my ands from my cold, dead hands.

  43. I moved to a new school in third grade, with a somewhat different curriculum emphasis than my earlier shcools. My math teacher that year really pushed us not to use the “and” between the hundreds and tens digits of numbers. I refused to do it except when being quizzed in class, and I later noted that the teacher herself normally used the “and” and omitted it only when she was teaching us the supposedly proper nomenclature of numbers.

  44. Such is prescriptivism: even the prescribers can’t stick to it. I can see why we don’t say “five thousand and two hundred and eighty”, because the “five thousand and two” garden-paths the listener. As for “and” avoidance, that too is supposed to prevent a garden path, not with decimals but with mixed fractions, aka compound numbers: “one hundred and five eighths”, for example, as opposed to “one hundred five”. But I call it foolishness: a number is over when it’s over.

    In Arabic, long numbers are spoken “seven and thirty and four hundred and one thousand”, which is written 7-3-4-1 and appears (right to left) as 1437. But in other right-to-left languages, even those written in the Arabic script like Persian and Urdu, numbers are spoken big-endian as in English and are written from left to right, so the writing hand has to jump to the right, leaving enough room for the digits to be written, and then the hand jumps to the right of the digits to resume writing letters. For this reason as well as differences of shape, Unicode provides two different sets of Arabic-script digits.


    ٠ ١ ٢ ٣ ٤ ٥ ٦ ٧ ٨ ٩

    ۰ ۱ ۲ ۳ ۴ ۵ ۶ ۷ ۸ ۹.

  45. George Gibbard says

    Actually in Arabic one says “(a) thousand and four of (a) hundred and seven and thirty”; the written ordering convention which does not match the spoken expression was borrowed along with the numeral symbols from India. Wikipedia is quite good on the grammar of numerals in Arabic:

  46. George Gibbard says

    Prior to introduction of Indian numerals, the Arabs used the letters of the alphabet to write numbers, in the same way as in Hebrew or Ancient Greek: that is, the first 9 letters of the Arabic alphabet (in the older, Aramaic order) represented numerals 1-9, the next 9 represented the tens 10-90 (with sīn replacing the obsolete samekh in the most common variant), the last four Aramaic-derived letters were 100-400 and the remaining hundreds and 1000 were represented by the new letters formed by adding dots to represent interdental, uvular and lateral (ɮʶ) fricatives — usually, in the order in which these sounds appear in the newer alphabetical order.

    Old (Aramaic) order: alif (1), bāʼ (2), jīm (3), dāl (4) …
    Newer order: alif (1), bāʼ (2), tāʼ (400), ṯāʼ (500), jīm (3), ḥāʼ (8), ḵāʼ (600), dāl (4), ḏāl (700) … (where bāʼ, tāʼ and ṯāʼ have the same shape and differ only by dots, as do jīm : ḥāʼ : ḵāʼ and also dāl : ḏāl.)

    Now it turns out that while ‘one hundred fifty-four’ is miʼatun wa-ʼarbaʻatun wa-ḵamsūna, ‘(a) hundred and four and fifty’, the system already was to write the hundreds before the tens before the ones: قند qnd, which as you can see ends with the four (dāl).

    Some peculiarities are that ‘2’ (bāʼ) and ‘3’ (jīm) were written without the normal under-dot, while jīm lacked the usual tail, in order to distinguish it from ‘8’ (ḥāʼ). Information from:

  47. “For this reason [an infelicity of hand-writing numbers in, especially, Persian and Urdu] as well as differences of shape, Unicode provides two different sets of Arabic-script digits.”

    In my recent experience in Iran, both sets of digits are used indifferently, with maybe a slight bias towards the Persian digits in official material and a slight bias away from them in non-official material. Here’s a taxi in Tabriz, note the ‘Arabic’ four and the ‘Persian’ six in the stencilled taxi plate number above the right rear light. Mixing the two in one group of digits is a bit out of the ordinary, which is why I have the photo handy, but both are used widely.

    If Iranian practice reflects practice elsewhere that the script is used, Unicode was inconsistent with its own policies in encoding them twice, the difference between ٤ and ۴ is no more significant than that between 7 with a stroke and 7 without.

  48. Shape was secondary: it was the directional properties that were critical. In the comment above, I entered the digits in the order 0-9 in both cases, but they appeared right-to-left for the Arabic, left-to-right for the Indic-Arabic set. A single character cannot have both LTR and RTL directionality. For the same reason, the mathematical aleph is a distinct character from the Hebrew one, even though they have identical glyphs: you want ℵ₀ and not א₀ to appear when you type “aleph null”.

Speak Your Mind