> My grandfather was an accountant who preferred the slide rule to a calculator.
This is hard for me to understand. Slide rules are inherently imprecise, in the same way floating point calculations are imprecise but much moreso. I'd expect an accountant to prefer integer (fixed point) math for the same reason banks do now.
Yes, they are imprecise. In general, you're limited to just three, maybe four digits regardless of the actual magnitude of the numbers and you have to keep the powers of ten in your head. Honestly, though, once you've used it enough, it isn't that hard to do. Besides, in most calculations, you don't need precision out to the 12th digit. After all, skyscrapers, bridges, and ocean liners have been built using slide rules for a long time and they, mostly, have worked just fine.
Imprecision is fine for engineering and can be compensated for. But this is not true for managing money, which is why floating point has historically not been used in accounting applications.
In some sense. A way to teach how a slide rule works is by using two rulers to create a slide rule that does addition. Most pupils will understand why that works. The jump from there to really understanding why logarithmic scales work is a lot harder for most, but I’m sure that intermediate step helps some pupils.
Oh, come on. I'm not talking about a home-brew slide rule or a virtual slide rule simulation. Obviously it is possible to make those. I'm talking about an actual manufactured slide rule sold as a product and used in a working environment to perform addition.
Most slide rules have a linear scale on them, but it is intended to be used along with a logarithmic scale for working with exponential/logarithm functions.
In general slide rules aren’t used for addition because 3-digit addition is very fast to compute mentally or using pen and paper.
I used to collect slide rules. I have never seen one with a linear scale, and I can't find any evidence on the web that any such slide rule was ever manufactured. So if you can provide evidence that you possessed such a slide rule, that would be of significant historical interest.
Isn't the "L" scale on a slide rule linear? See the photo at the top of the Wikipedia page [1] for example. (I may be confused here, since you know slide rules better than me.)
Sure, but to do addition you need two linear scales, one on the fixed part and one on the slider. The L scale is for computing logarithms, not for doing additions.
If the L scale is on the slider, and you’re not too much concerned accuracy (the procedure does 5 ‘align scales’ operations before the final ‘read the result’) one linear scale is sufficient, if you use the hairline on the cursor as a kind of memory:
”Example: calculate 0.23 + 0.45
- "Reset" the rule so that all the scales are lined up.
- Move the cursor to 0.23 on the L scale.
- Move the leftmost 0 on the L scale to the hairline.
- Move the cursor to 0.45 on the L scale.
- Reset the rule again so that all the scales are lined up.
- The cursor should now be at 0.68 on the L scale, which is the sum of 0.23 + 0.45.”
Using a sledge hammer, break the slide rule into N pieces, taking care to insure that N is greater than the sum you wish to compute. Now to compute the sum of A and B, count out A pieces into a paper bag, then count out B pieces into the same paper bag. Now empty the bag onto a clean workspace and count the total number of pieces to produce the result.
I checked, and sure enough you're correct. I could have sworn that I'd had two linear scales, but … I was completely wrong. Been decades since I used it, and the memories obviously faded.
This is hard for me to understand. Slide rules are inherently imprecise, in the same way floating point calculations are imprecise but much moreso. I'd expect an accountant to prefer integer (fixed point) math for the same reason banks do now.