## Wolfram Alpha and Babylonian

EDIT: Wolfram Alpha has been updated to fix the problem mentioned below.

One of the more curious features of Wolfram Alpha is if you type a single number, in addition to giving you trivia it will (if you so choose) render historical versions of that number. Here’s some selections in Babylonian numerals:

Here are what Babylonian numerals are supposed to look like. I originally thought Wolfram Alpha wasn’t using base 60 like it’s supposed to be, but I realized later they are just separating out the renderings (like 38 above) so they look like separate symbols in base 10. However, there’s still weird errors going on (like with 10 above, which is clearly not the right symbol) but I have no idea what’s the pattern to the errors. Anyone know?

### 13 Responses

1. It looks to me that they’re using a mixed-radius system in which the overlapped-ten symbol is a zero. So 10 = 1×10 + 0x1, 38 = 3×10 + 8×1, 61 = 1×60 + 0x10 + 1×1.

Of course this is wrong. The Babylonian notation is sexagesimal, not mixed-radius, with each sexagesimal digit written as a combination of 10s and 1s, so the two digits in 61 are both 1s with no zero separating them, and similarly there is no zero in 10.

2. You seem to be right except:

The “38” was spaced as in the picture above, suggesting it is intended to be a single digit. Perhaps some funky coding behind the scenes ended up giving them both systems at the same time?

3. If you walpha “0 babylonian”, you’ll get that overlapped ten symbol all by itself.

It appears they want to use it as a placeholder for both an entire “place” (as in 60), but also as a placeholder for unused ones or tens within a place (as in 61 above, or e.g. 80)

Curiously, it appears that walpha fails to give output for n babylonian, if n>3599.

4. Oh, and if you want to see what actual Babylonian numbers look like, beyond the somewhat cleaned-up image in the Wikipedia article you link to, I think Plimpton 322 is a good example.

5. Cool that it gives the Babylonian, Mayan, Roman and Greek as references. Not sure about calling them the “antique” number systems though. 🙂

6. […] Number Warrior, Jason Dyer, writes about an apparent bug in Wolfram Alpha. How can we trust Wolfram to get complex integrals right if they can’t even […]

7. […] easter eggs helpful ~ Crossword puzzle solver in need of instructions ~ A new kind of wiki wrong ~ Babylonian numerals imitated ~ Harvey|Omega not ready for this one ~ A tough […]

8. […] Incidentally, if you walpha a number, Wolfram Alpha will give it to you in Babylonian symbols.  Here’s  34.  But watch out: it’s sometimes wrong. […]

9. […] and Babylonian has been fixed Posted on October 7, 2009 by Jason Dyer Prior post update: Wolfram Alpha’s issues with Babylonian have been […]

10. WolframAlpha (Wolfa.com) has thankfully emerged…

I was getting tired of typing the long name.

11. Note that wolfa.com is a FAKE domain, not registered to Wolfram|Alpha LLC and likely to become an attack vector for malware in the future.

12. Dear Mr. Fake FBI agent,

Clearly you are not with the FBI or you wouldn’t make an ignorant allegation without proof. The owner of wolfa.com works for Wolfram and set up the domain as a courtesy to users of the service. The site will remain as it is now… a secondary URL to wolframalpha.com for convenience.

13. good to know – I heard it was just set up by some wolframalpha advocate to help boost popularity – either way its great and i am happy for the tinier name and wolframalpha