ADVERTISEMENT
  About the SA Blog Network













Roots of Unity

Roots of Unity


Mathematics: learning it, doing it, celebrating it.
Roots of Unity Home

10 Secret Trig Functions Your Math Teachers Never Taught You

The views expressed are those of the author and are not necessarily those of Scientific American.


Email   PrintPrint



On Monday, the Onion reported that the “Nation’s math teachers introduce 27 new trig functions.” It’s a funny read. The gamsin, negtan, and cosvnx from the Onion article are fictional, but the piece has a kernel of truth: there are 10 secret trig functions you’ve never heard of, and they have delightful names like “haversine” and “exsecant.”

A diagram with a unit circle and more trig functions than you can shake a stick at. (It's well known that you can shake a stick at a maximum of 8 trig functions.) The familiar sine, cosine, and tangent are in red, blue, and, well, tan, respectively. The versine is in green next to the cosine, and the exsecant is in pink to the right of the versine. Excosecant and coversine are also in the image. Not pictured: vercosine, covercosine, and haver-anything. Image: Tttrung and Steven G. Johnson, via Wikimedia Commons.

Whether you want to torture students with them or drop them into conversation to make yourself sound erudite and/or insufferable, here are the definitions of all the “lost trig functions” I found in my exhaustive research of original historical texts Wikipedia told me about.

Versine: versin(θ)=1-cos(θ)
Vercosine: vercosin(θ)=1+cos(θ)
Coversine: coversin(θ)=1-sin(θ)
Covercosine: covercosine(θ)=1+sin(θ)
Haversine: haversin(θ)=versin(θ)/2
Havercosine: havercosin(θ)=vercosin(θ)/2
Hacoversine: hacoversin(θ)=coversin(θ)/2
Hacovercosine: hacovercosin(θ)=covercosin(θ)/2
Exsecant: exsec(θ)=sec(θ)-1
Excosecant: excsc(θ)=csc(θ)-1

I must admit I was a bit disappointed when I looked these up. They’re all just simple combinations of dear old sine and cosine. Why did they even get names?! From a time and place where I can sit on my couch and find the sine of any angle correct to 100 decimal places nearly instantaneously using an online calculator, the versine is unnecessary. But these seemingly superfluous functions filled needs in a pre-calculator world.

Numberphile recently posted a video about Log Tables, which explains how people used logarithms to multiply big numbers in the dark pre-calculator days. First, a refresher on logarithms. The equation logbx=y means that by=x. For example, 102=100 so log10100=2. One handy fact about logarithms is that logb(c×d)=logbc+logbd. In other words, logarithms make multiplication into addition. If you wanted to multiply two numbers together using a log table, you would look up the logarithm of  both numbers and then add the logarithms together. Then you’d use your log table to find out which number had that logarithm, and that was your answer. It sounds cumbersome now, but doing multiplication by hand requires a lot more operations than addition does. When each operation takes a nontrivial amount of time (and is prone to a nontrivial amount of error), a procedure that lets you convert multiplication into addition is a real time-saver, and it can help increase accuracy.

The secret trig functions, like logarithms, made computations easier. Versine and haversine were used the most often. Near the angle θ=0, cos(θ) is very close to 1. If you were doing a computation that had 1-cos(θ) in it, your computation might be ruined if your cosine table didn’t have enough significant figures. To illustrate, the cosine of 5 degrees is 0.996194698, and the cosine of 1 degree is 0.999847695. The difference cos(1°)-cos(5°) is 0.003652997. If you had three significant figures in your cosine table, you would only get 1 significant figure of precision in your answer, due to the leading zeroes in the difference. And a table with only three significant figures of precision would not be able to distinguish between 0 degree and 1 degree angles. In many cases, this wouldn’t matter, but it could be a problem if the errors built up over the course of a computation.

The bonus trig functions also have the advantage that they are never negative. Versine ranges between 0 and 2, so if you are using log tables to multiply with a versine, you don’t have to worry about the fact that the logarithm is not defined for negative numbers. (It is not defined for 0 either, but that is an easy case to deal with.) Another advantage to the versine and haversine is that they can keep you from having to square something. A little bit of trigonometric wizardry (a.k.a. memorization of one of the endless list of trig formulas you learned in high school) shows that 1-cos(θ)=2sin2(θ/2). So the haversine is just sin2(θ/2). Likewise, the havercosine is cos2(θ/2). If you have a computation involving the square of sine or cosine, you can use a haversine or havercosine table and not have to square or take square roots.

A diagram showing the sine, cosine, and versine of an angle. Image: Qef and Steven G. Johnson, via Wikimedia Commons.

The versine is a fairly obvious trig function to define and seems to have been used as far back as 400 CE in India. But the haversine may have been more important in more recent history, when it was used in navigation. The haversine formula is a very accurate way of computing distances between two points on the surface of a sphere using the latitude and longitude of the two points. The haversine formula is a re-formulation of the spherical law of cosines, but the formulation in terms of haversines is more useful for small angles and distances. (On the other hand, the haversine formula does not do a very good job with angles that are close to 90 degrees, but the spherical law of cosines handles those well.) The haversine formula could yield accurate results without requiring the computationally expensive operations of squares and square roots. As recently as 1984, the amateur astronomy magazine Sky & Telescope was singing the praises of the haversine formula, which is not only useful for terrestrial navigation but also for celestial calculations. For more on the haversine formula and computing distances on a sphere, check out this archived copy of a census bureau page or this Ask Dr. Math article.

I don’t have much information about the history of the other trig functions on the list. All of them could make computations more accurate near certain angles, but I don’t know which ones were commonly used and which ones were named* analogously to other functions but rarely actually used. I’m curious about this, if anyone knows more about the subject.

When the Onion imitates real life, it’s usually tragic. But in the case of secret trig functions, the kernel of truth in the Onion didn’t make me sad. We’re very lucky now that we can multiply, square, and take square roots so easily, and our calculators can store precise information about the sines, cosines, and tangents of angles, but before we could do that, we figured out a work-around in the form of a ridiculous number of trig functions. It’s easy forget that the people who defined them were not sadistic math teachers who want people to memorize weird functions for no reason. These functions actually made computations quicker and less error-prone. Now that computers are so powerful, the haversine has gone the way of the floppy disc. But I think we can all agree that it should come back, if only for the “awesome” joke I came up with as I was falling asleep last night: Haversine? I don’t even know ‘er!

*I’d like to take a little digression to the world of mathematical prefixes here, but it might not be for everyone. You’ve been warned.

In the table of secret trig functions, “ha” clearly means half; the value of haversine is half of the value of versine, for example. “Co” means taking the same function but with the complementary angle. (Complementary angles add up to 90 degrees. In a right triangle, the two non-right angles are complementary.) For instance, the cosine of an angle is also the sine of the complementary angle. Likewise, the coversine is the versine of the complementary angle, as you can see in light blue above one of the red sines in the diagram at the top of the post.

The one bonus trig function that confuses me a little bit is the vercosine. If the “co” in that definition meant the complementary angle, then vercosine would be the same as coversine, which it isn’t. Instead, the vercosine is the versine of the supplementary angle (supplementary angles add up to 180 degrees), not the complementary one. In addition to the definitions as 1-cos(θ) and 1+cos(θ), the versine and vercosine can be defined as versin(θ)=2sin2(θ/2) and vercos(θ)=2cos2(θ/2). In the case of versine, I believe the definition involving cos(θ) is older than the definition involving sine squared. My guess is that vercosine was a later term, an analogy of the square of sine definition of versine using cosine instead. If you’re a trigonometry history buff and you have more information, please let me know! In any case, the table of super-secret bonus trig functions is a fun exercise in figuring out what prefixes mean.

Evelyn Lamb About the Author: Evelyn Lamb is a postdoc at the University of Utah. She writes about mathematics and other cool stuff. Follow on Twitter @evelynjlamb.

The views expressed are those of the author and are not necessarily those of Scientific American.





Rights & Permissions

Comments 18 Comments

Add Comment
  1. 1. Agesilaus 4:03 pm 09/12/2013

    Most of these were used quite a bit in Calculus, at least when I took it back in the 1970′s.

    Link to this
  2. 2. m 8:39 pm 09/12/2013

    Back in my day, when we used quills and ink blotting paper… we really didn’t care what we wrote as we made a mess either way.

    Link to this
  3. 3. Andragogue 12:37 am 09/13/2013

    I did it all with a slide rule to earn my physics degree in 1971. I used to think it was fun and a challenge, but I can hardly remember any trig any more; nonetheless, I can see where a little knowledge of these extra trig functions could have made solving a big, hairy integral somewhat more straightforward.
    (P.S. Loved your jokes, too!)

    Link to this
  4. 4. Michael Hardy 12:38 am 09/13/2013

    Any account of things like this is incomplete without mentioning Ptolemy’s table of chords, tabulated in immense detail during the second century:

    http://en.wikipedia.org/wiki/Ptolemy%27s_table_of_chords

    Link to this
  5. 5. Layer_8 2:03 am 09/13/2013

    Some more: sinc(t), Si(t), si(t), sinh(t)…

    Basically all you need is log(z) for z being a complex number :-)

    Link to this
  6. 6. Dave Bell 4:16 am 09/13/2013

    I first came across these trig functions in connection with navigation. There’s a lot of the skills involved in that which, with GPS, are fading from the world.

    I am part of the last generation to be exposed to slide rules at school. I never saw one with these functions, but maybe a related effect was the need to do rough mental arithmetic to fix the decimal point in a calculation. I hadn’t thought of the limits of the log tables, but it is obvious when you see it.

    As for the log of a negative number, if you can’t multiply by -1 without log tables, there’s something wrong.

    We forget so much.

    Link to this
  7. 7. elderlybloke 4:43 am 09/13/2013

    I spent a few “happy” years doing survey calcs.with my Boss sitting on Hills , finding the Log values him to do adding and subtracting , and eventually getting information we required.
    Then , Happy Days, HP produced a Pocket Calculator for Surveyers.
    The best invention since Sliced Bread.

    The speed and accuracy improvement was well worth the apparent high cost.

    Link to this
  8. 8. einsteinscat 10:02 am 09/13/2013

    Always was poor at arithmetic because it was a bore and just straight memorization. Always had poor to honestly vicious math teachers, so I learned to either believe I had no math ability or they were prejudiced.
    Later discovered Math as a concise way of saying something, that anyone else knowing the language could unravel, disprove or elaborate upon.
    But, the natural connection of sine cosine and graphics with Natural cycles is what always blew my mind away. Is our world a simulation on a Universal computer. Nobody yet can say

    Link to this
  9. 9. johnepopp 6:18 pm 09/13/2013

    These ‘new’ trig functions are nicely explained because the language is logical and easily remembered. The opposite is true of logarithms. The meaning of this function was always a struggle for me to remember until one day a simpler alternative occurred to me. If 10 to the power (or exponent) 2 is 100, why don’t we say 2 is the power (or exp) of the base 10 of the number 100, i.e. pow(subscript 10)100 = 2 . There is no need for the word logarithm. My alternative describes its meaning.

    Link to this
  10. 10. Layer_8 8:07 am 09/14/2013

    @9. You’re very right. Logarithm IS the power. That’s what we were taught at school. And you need only 5 Numbers: 0 = 0
    1 = 1
    i² = -1
    e = 2.7182818…
    pi = 3.1415926…

    log_10(x) = log_e(x)/log_e(10)
    log_2(x) = log_e(x)/log_e(2)

    x = exp(log_e(x))
    cosh(x) = {exp(x) + exp(x)}/2
    sinh(x) = {exp(x) – exp(x)}/2
    tanh(x) = sinh(x)/cosh(x)
    cos(x) = {exp(ix) + exp(-ix)}/2
    sin(x) = {exp(ix) – exp(-ix)}/2
    tan(x) = sin(x)/cos(x)

    Finally exp(i*pi) + 1 = 0

    Everything comes out of logarithms.

    Link to this
  11. 11. spike7638 11:47 am 09/14/2013

    When a friend sent me a link talking about “trig functions you’ve never heard of,” I was sorry to see that they were ones I *had* heard of…but it’s an opportunity to share another — the gudermanian, whose inverse has, as its integral, ln| sec x + tan x |.

    Spivak’s wonderful calculus book has a problem putting this in a historical context; in part (a) the reader finds the log | sec x + tan x | integral for secant, but part (b) suggests another approach:

    (b) [do the integration] By using the substitution t = tan(x/2). [...] we obtain int sec(x) dx = log(tan(x/2 + pi/4)). This last expression was actually the first one discovered, and was due not to any mathematician’s cleverness, but to a curious historical accident: in 1599 Write computed nautical tables that amounted to definite integrals of sec. When the first tables of the logarithms of tangents were produced, the correspondence between the two tables was immediately noticed (but remained unexplained until the invention of calculus).

    Ever since I read that problem, in 1970, I’ve wondered what sort of person “immediately notices” that two completely different tables actually have the same contents. :)

    Link to this
  12. 12. spike7638 2:11 pm 09/14/2013

    Oops. Spelling error. It’s actually “gudermannian”. Apologies.

    Link to this
  13. 13. wrbilledwards 6:06 pm 09/14/2013

    Prof Gudermann must be very hurt, wherever he is!

    Link to this
  14. 14. wrbilledwards 6:10 pm 09/14/2013

    Thanks, Dr. Lamb, we needed this. Trig has not made me smile in a _very_ long time!

    Link to this
  15. 15. magdrop 8:53 am 09/24/2013

    This haversine vs. cosine question is really bugging me now. The new consensus among all of the keyboard warriors seems to be that the two are mathematically identical and yet give different results only due to rounding or precision error.

    I’m pretty sure that’s wrong, but I’m not positive.

    For some reason, the two give different results when mapped onto the complex plane:

    http://mathworld.wolfram.com/images/interactive/CosReImAbs.gif

    http://mathworld.wolfram.com/Cosine.html

    http://mathworld.wolfram.com/images/eps-gif/HaversineReIm_850.gif

    http://mathworld.wolfram.com/images/eps-gif/HaversineContours_850.gif

    http://mathworld.wolfram.com/Haversine.html

    Clearly, they might not be identical mathematically. I honestly don’t think that they are.

    I think they are each derived differently, and the cosine behaves real squirrelly whenever the angle approaches zero degrees.

    Weird, huh?

    Link to this
  16. 16. aerobert 2:47 pm 09/24/2013

    The diagram prompted reminiscence of the connection between names of the lines and the functions. Ask most people today and they give arithmetic definitions — the wikipedia page (Unit_circle) similarly gets straight into Cartesian coordinates — I much prefer terms based on the geometry of an arc.

    Haversine is pretty good for interpolating when you don’t want the crudeness of a linear slope. It’s used by the controller on our mechanical test rig to avoid jerks! — geeks are harder to avoid in an engineering lab ;-)

    Link to this
  17. 17. AlejandraMarroquin 7:15 pm 10/2/2013

    Could you please check colors relationships with trig functions in first diagram.
    Thanks

    Link to this
  18. 18. Evelyn Lamb in reply to Evelyn Lamb 11:27 pm 10/2/2013

    Thanks, AlejandraMarroquin. The caption has been corrected.

    Link to this

Add a Comment
You must sign in or register as a ScientificAmerican.com member to submit a comment.

More from Scientific American

Scientific American Dinosaurs

Get Total Access to our Digital Anthology

1,200 Articles

Order Now - Just $39! >

X

Email this Article

X