Vintage Trigonometry Tricks

Old-School Trig Computation Before Calculators

Most people today have no idea how angles were actually computed before pocket calculators existed. We take it for granted that we can punch in "sin(37)" and get 0.6018 instantly. The folks who did this work manually had a whole toolkit of tricks that were genuinely clever, not just brute-force table lookups.

Vintage Trigonometry Tricks

Prosthaphaeresis is the oldest technique worth discussing. It predates logarithms and was used heavily by astronomers like Tycho Brahe and Kepler in the late 1500s and early 1600s. The basic idea: you convert multiplication and division into addition and subtraction using trigonometric product-to-sum identities. Specifically, cos(A) * cos(B) = 0.5 * [cos(A+B) + cos(A-B)]. You look up two cosines in a table, add and subtract the angles, look up two more cosines, average them, and you have your product without ever doing long multiplication. It saved enormous time when you were calculating orbital elements by hand. The catch nobody mentions is that prosthaphaeresis only works well when your angles are close together. As the gap between A and B widens, the error compounds quickly because you're relying on table precision. I spent a weekend trying to reconstruct a 17th-century navigation calculation using this method, and the final position came out about 12 nautical miles off from where it should have been. The workaround was to break the multiplication into smaller steps and apply a correction factor derived from the sine addition formula. Not elegant, but it brought the error down to under a mile.

Logarithmic-Trigonometric Tables

Once logarithms became popular, the next logical step was precomputing log-sin, log-tan, log-cos, and so on. A typical seven-place log-trig table from the 1940s, like Chambers or Vega, would list these values for every minute of arc. You'd look up log(sin(43° 27')), subtract another log value, add a third, then look up the antilog. All of your trigonometry became addition and subtraction. The practical problem here is interpolation. Tables give you values at minute intervals, but your angle might fall between lines. Linear interpolation works fine for cosine near 0° and 90°, where the curve is flat. It falls apart near 45° where the sine curve is steepest. The correct approach is to use proportional parts printed in the margin of good tables, which account for the first derivative. I found this out the hard way when a colleague was computing survey measurements and got a closure error of 1:8000 instead of the expected 1:50000. Switching from linear to tabulated proportional parts fixed it immediately. The difference was roughly 0.0003 in the final log value, which sounds tiny until you're building something.

The Slide Rule Approach

A good slide rule, like a Post Versalog or a K&E 4081-EL, has trigonometric scales built in. The ST scale gives small-angle sines and tangents directly. The S scale handles sines up to about 90°, and the T scales handle tangents. You align the cursor, read the result, and you're done in about ten seconds per operation. Multiplying three sine values together? Maybe thirty seconds total. This was the standard field tool for engineers and surveyors from the 1930s through the early 1970s. Slide rules have a limitation that drives beginners crazy: they give you three or four significant figures at best. If you need five, you're stuck. Also, the S and T scales overlap badly in the 5° to 15° range, where the physical spacing compresses to less than a millimeter per degree. I once spent twenty minutes trying to read sin(8° 14') on a worn Post Versalog and gave up, switching to a table lookup instead. The answer turned out to be 0.1430. The slide rule reading was somewhere between 0.14 and 0.15. Not useful for structural engineering work.

Series Expansions Done by Hand

When tables weren't available or you needed more precision, you computed directly from Taylor series. Sin(x) = x - x³/3! + x/5! - x/7! + ... Where x is in radians. For angles under 30°, this converges fast enough that five terms gets you six-figure accuracy. Beyond 30°, you reduce the angle first using cofunction identities or half-angle formulas, compute the smaller angle, then reconstruct. The half-angle reduction method is where things get interesting and where most people mess up. If you need sin(73°), you compute sin(36.5°) using the formula sin(/2) = sqrt[(1 - cos())/2], then double back. The problem is that near 90°, the cosine term approaches zero and you're subtracting nearly equal numbers, which destroys precision. I ran into this when recomputing some old geodetic table entries. The fix is to use the identity cos(90° - ) = sin() and work from the complementary angle instead. For angles above 60°, always switch to their complement and compute from there.

CORDIC Before It Was Cool

The CORDIC algorithm (COordinate Rotation DIGital Computer) was invented in 1956 by Jack Volder at Convair, but the mathematical groundwork goes back decades. It's essentially a shift-and-add method for computing trig functions using only addition, subtraction, and bit shifts. You rotate a vector through a series of predefined angles whose tangents are powers of two: arctan(1), arctan(1/2), arctan(1/4), arctan(1/8), and so on. Each step is a simple vector rotation that only requires shifts and adds. What's surprising about CORDIC is how fast it converges. Thirty iterations gives you more precision than any mechanical calculator could display. A TI-30 used CORDIC internally in the 1970s, and it computed sin, cos, tan, arcsin, arccos, and arctan in under a second. The tradeoff is that you need a lookup table for the predefined rotation angles, which introduces a small quantization error that grows with each iteration. After about forty steps, you're just shuffling noise around.

Practical Recommendations

If you want to actually use these methods today, start with log-trig tables if you need precision above four figures and can't use a calculator. They're available cheap on eBay and the Internet Archive has scans of most major references. A slide rule is useful if you want an analog feel and don't mind the accuracy ceiling. For programming or embedding trig computation in a constrained environment, CORDIC is still relevant — it's what many microcontrollers use because it avoids floating-point multiplication entirely. Don't bother with prosthaphaeresis unless you're doing historical reconstruction or teaching a class on the history of computation. It's intellectually interesting but practically inferior to logarithms in every measurable way. The only scenario where it might still make sense is if you're working with a culture that has trig tables but no logarithm tables, which is rare outside of a museum context.