Hacker News Reader

Comment by jacquesm | Hacker News Reader

Coincidentally I'm making one right now. I never knew about Walter and his passion but I totally get it, sliderules have a magic all their own. Ostensibly just a couple of passive pieces of material but the power of logarithms turns them into universal calculators. A radio amateur that I knew was faster with his slide rule than I ever was with a calculator! The calculator produced more digits but most of those were useless.

View on HN

> but most of those were useless.

Has anyone ever made a calculator that manages significant digits ?

Edit: I mean, if (e.g.) both inputs to a binary operation have 5 digits of which only 3 are significant, then does said calculator grok this.

HP calculators will only display the number of digits you specify and there is an RND function which will round the number to that number of digits.

While it doesn't fulfill the criteria of "managing significant digits" (you still have to know at each point how many significant digits you have, which isn't obvious with calculations like logarithms), it does make things significantly (ahem) easier in most common cases.

Hehe, nice pun. :)

My high school Casio had eleven significant digits. You just wouldn't know it unless you read the manual.

How can a calculator have an innate number of sigfigs? Significant figures depend on on the number of digits of an input number

No, it depends on the significant digits of the input number. And more often than not there are very few of those, even though it may look like there are a lot. And your average calculator has no idea about how to deal with this properly. A typical measurement (say, temperature) has two significant digits at best, three if you're very precise and have a calibrated thermometer operating in the sweet spot of its scale. A measurement of time is hard to achieve to better than 10/th of a second without special preparation and as the interval you are measuring gets longer so will the error (because they tend to accumulate). A measurement of weight is so tricky that even three or four successive measurements are going to give you different results. Modern personal weight scales fake this by freezing the digital output at some arbitrary value and if you get off the scale and back on again it doesn't do another measurement but it simply displays the stored value to give you the illusion of precision where there really isn't any, another measurement might be off by a few hundred grams.

So for most problems that do not involve money but real world data the number of significant digits is very low. Two, three, maybe. Four in very rare cases. And when you start combining them you're quickly into 'order of magnitude' territory rather than anything precise.

Right, that's what I was saying. Which is why I was wondering about the claim that I'm replying to saying that a calculator itself has a specific number of significant digits independent of the input numbers

I think what they're getting at is that it has 11 digits internally but only displays the significant ones even though it has more 'under the hood'.

That's super cool! What are you making it out of?

I always wondered how they made the markings on the high quality slide rules of yore. Were they engraved? Were there machines that would scribe them? Now we might laser engrave, but that feels too recent. Maybe some sort of acid etching? I don't know!

For now out of wood, it will be 'oversized' so I get enough accuracy because I can only inscribe 0.4 mm lines, though I have an idea on how to improve on that for just this job.

The ones that I used to have were plastic and the lettering seemed to have been superimposed on top of the base material, Pickett's were definitely printed, not engraved. But older ones were engraved and I'm super curious about the process they used, I'm imagining a reversed pantograph or something like that.

Reply on news.ycombinator.com