Hacker News Reader

Comment by chrismorgan | Hacker News Reader

Feels strange to have single decimal places in your inch figures. I’d have expected something like eighths, sixteenths and thirty-seconds of inches, even though the decimal representation of them is unwieldy (1⁄32″ becomes 0.03125in). That also helps you avoid fractional pixels, which is worth doing if convenient: 1⁄32″ is 3px, because 1in is defined as 96px.

(As an Australian, the main place I’ve ever seen fractions of inches is old tools like spanners from before the adoption of the metric system, and that’s all fractions with powers-of-two denominators. But maybe Americans use decimal fractions of inches? I suppose I have seen laptop screens described in that way, e.g. 13.1″, 15.4″ and 15.6″, though people typically truncate to the inch.)

View on HN

All small measurements in the US are done in power-of-two fractions of an inch. In fact, much of the initial opposition to the metric system here was led by the construction industry, who has found from experience that 1/2" 1/4" 1/8" 1/16" etc. measurements are too convenient to ever give up.

How this that convenient when you then have to use something like 13/16?

Just look at the standard wrench sizes (https://www.cnm.edu/programs-of-study/programs-a-z/automotiv...):

* Standard Combination Wrenches: 1/4, 5/16, 11/32, 3/8, 7/16, 1/2, 9/16, 5/8, 11/16, 3/4, 13/16, 7/8, 15/16, 1.

* Metric Combination Wrenches: 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19. Measured in millimeters.

Thanks for this illustration.

Perfectly convenient.

The standard system excels when doing actual work. If you have a rod with 1" diameter and you need to take it down to 13/16", you can use a lathe to make 3/32" cut. Or if you need to divide up 13/16 into 4 parts, for example, 13/64.

The metric system's strength is teaching in elementary schools where skills with fractions are not as strong.

Not exactly. Things are done in decimal fractions until you get down to around 1/64", but any farther and they're done in thou (1/1000" aka mils) or tenths (1/10000"). There's overlap, machinists will use thou up to and sometimes beyond a full inch.

Rulers, drill bits, and wrenches typically use fractions of an inch where the denominator is a power of two, but circuit board design often uses decimal fractions of an inch. The holes in a breadboard are commonly .1 inches apart.

Then the whole mils thing...

Routing a PCB is like constantly flipping your ruler around: Old, classic through-hole parts are in inches/mills, while new, modern surface-mount parts are in millimeters. All electrical aspects (traces, spacing, lengths) are measured in inches/mills, and often the mechanical aspects are measured in millimeters (if the enclosure is not made in the U.S.). And if you use the wrong system, you engineering tables and formula stops working, the whole thing is out-of-context (e.g. measuring the gap in a differential pair in millimeters, the number makes no sense), and an integer becomes a float with three significant digits after the decimal point.

So here is the joke: The electronics industry is moving towards the metric system, one mill at a time. At least CAD programs have a button to switch between the two instantly.

Engineering and machining specifications are usually in 0.001” and referred to as thousandth’s or “thou’s”. Tools like wrenches or drill bits use fractions in the form x+1/2^n like 2 1/2 == 2.5”. It goes down to 1/64, I think.

They also speak of tenths which are 0.0001" units, i.e. "tenth of a thou". This jargon tends to confuse those who are new to it, but has been around for over a century.

It's a really well-sized unit for that kind of work. I know exactly what a thou is and how to keep things in spec, but I'm lost using mm.

Thous, and tenths even more so, are more akin to microns in metric.

Since several of the lengths are in fifths of an inch, I thought it was a secret desire for the metric system.

We usually use fractions, but sometimes we're forced to use decimals, usually when related to computers, since the usual fractions don't translate well to decimal.

Wait, what?

You're saying that Americans preferentially use base-2 fractions (nearly) everywhere except on computers; on computers you preferentially use base-10, even though said computers are going to store the values as base-2?

That's the funniest thing I've heard all week.

A feature of some English measures is unbiased division into three equal parts, e.g. Three teaspoons to the tablespoon, twelve inches to the foot, and twelve ounces to the pound (for one of several definitions of “ounce”).

Thirds don’t fit neatly into binary nor do they fit neatly into decimal. This is what the comment refers to.

> "on computers you preferentially use base-10, even though said computers are going to store the values as base-2?"

I don't understand this comment at all. Most people are not computer programmers and therefore have no insight into how computers represent numbers. They just populate text fields in Excel, or whatever, with numbers. It's easier to use decimal points to do that than fractions.

When I'm not using computers or calculators, I do prefer American-style fractions over decimal point representations. Fractions are a lot easier for me to manipulate in my head. If I'm out in my garage, upside down, underneath a car, and trying to decide which drill bit to use, I don't have convenient access to the calculator on my cell phone or a pencil and paper, and reliable mental math is important. The fractions are easier, so that's what I want.

You find it more difficult to choose between 1mm, 2mm, 3mm, 3.5mm, 4mm, 4.5mm, 5mm, 6mm, 7mm, 8mm drill bits than 1/16 5/64 3/32 7/64 1/8 9/64 5/32 1/4 7/32 3/16 inches?

What's 3/16" in 64ths, while under your car?

(Imperial numbers copied from a website selling a set, metric from my memory of my 10-bit set. I doubt these sets are similar.)

> What's 3/16" in 64ths, while under your car?

12. You just double it, twice.

I'm British, and generally partial to metric units, but power-of-two fractions actually seem quite nice to me.

I believe they're saying it's ironic that, despite the base-2 values being ideal for binary computers, users are having to use base-10 to enter them into the computer and hence getting none of the possible advantage. I don't believe they're making a value judgement.

Reply on news.ycombinator.com