1. Not finding help here? Sign up for a free 30min tutor trial with Chegg Tutors
    Dismiss Notice
Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Simple Significant Figure Problem

  1. Aug 25, 2016 #1
    1. The problem statement, all variables and given/known data
    Convert ##120. \frac{km}{h}## to ##\frac{m}{s}##

    2. Relevant equations
    ##1 km=10^3 m##
    ##1 hour = 3600 seconds##

    3. The attempt at a solution
    Not so much concerned with the solution, it's easy, as to what happened after the solution was found. So you get ##33.33\overline{33}##, after which the professor rounded to 33.3 citing significant figures. That makes no sense though, if we go back in the other direction we would see this new value is actually ##119.88\frac{km}{h}## which is more precise than the original value. If you used this value in another calculation it seems like it would defeat the entire purpose of doing the whole silly significant figure thing in the first place.

    Now you could just cite significant figures again and get back to 120., but if you wanted to convert the value in to use again at a later date you'd have rounding error. The chemistry lecture professor basically brushed me off when I brought it up. I was arguing that the entire ##33\frac{1}{3}## was significant.

    Who's right? Any insights here?
     
  2. jcsd
  3. Aug 25, 2016 #2

    haruspex

    User Avatar
    Science Advisor
    Homework Helper
    Gold Member
    2016 Award

    I believe the decimal point shown in "120." says the 0 is a significant figure (not sure how it works if it had been 1200 with three sig figs). This means you should take the exact value as being anywhere in the range 119.5 to 120.5. On dividing by 3.6 that gives 33.2 to almost 33.5, so even quoting 33.3 (implying 33.25 to 33.35) is overexact. But hey, there's only so much you can do without quoting specific bounds.
    Give your prof a break.
     
  4. Aug 25, 2016 #3

    David Lewis

    User Avatar
    Gold Member

    I believe you add one or two more significant figures when you run the computation backwards.
    (e.g. 33.333) That prevents imprecision from compounding.
    But you still round the final answer off to 3 significant figures when you're done.
     
    Last edited: Aug 25, 2016
  5. Aug 26, 2016 #4
    This ain't really helpful.

    Probably use scientific notation? I would guess.

    I should have said so, my bad, but the class generally assigns the last digit as the uncertain one and puts the error bounds at ##\pm 1##, so 119 to 121 km. That's another convention that doesn't seem very logical to me. If I were to use a measuring device with increments in which I would guess at a whole kilometer, I don't see how I could possibly be an entire kilometer off in either direction. This seems like it would true for any measurement that's around everyday units. ##\pm .5## makes a lot more sense. End rant.


    So on dividing by 3.6 we would get ##33+\frac{1}{18}## to ##33 + \frac{11}{18}##, so I just think we should keep the ##33 + \frac{1}{3}## and specify a new uncertainty of ##\pm\frac{5}{18}## to propagate both the precision and uncertainty correctly of the original measurement.

    Would this be right?

    It would be easier if the professor didn't say other things which made no sense, like accuracy of measuring devices isn't important, only precision is. She went on to say they pay 2000 dollars for each digit after the decimal, and I told her I would sell her a scale that had ##10^{whatever}## precision for 2 grand a digit, but wasn't accurate at all if that's what she thought. Maybe she brushed me off because I came across like a dick, but I only wanted to illustrate the silliness of the statement.

    Last semester the professor in the same department said similar things, but was actually worse. He gave me a F on one exam because I didn't round after each step in a calculation. I tried to explain that's not how it works, and eventually had to go to the dean. He hated me the rest of the course, possibly because I made him look like an idiot in front of the dean.

    I don't think he was an idiot, he had a Ph.D in chemistry, I just don't feel like they care or think we as students are too dumb to learn things the correct way. Maybe we are, but try me first before preaching something that seems error ridden.

    I just want to learn things the correct way, it's frustrating when none of my science courses can agree on how to handle the math in science. At least the physics courses generally treat this stuff better, it makes me question my desire to be a doctor and continue with this chemistry department. end second rant.
     
  6. Aug 26, 2016 #5

    haruspex

    User Avatar
    Science Advisor
    Homework Helper
    Gold Member
    2016 Award

    That prof gets an F from me. I regularly have occasion on these forums to point out that rounding prematurely leads to accumulation of errors. By the time you're done you might have no precision left.
    You should keep it in decimals. Taking the original to be +/-1 (ugh) that gives 33.3+/-0.3.
     
  7. Aug 26, 2016 #6
    Got it, thank you so much for the help and for letting me vent.
     
  8. Aug 27, 2016 #7
    That prof gets an F from me. I regularly have occasion on these forums to point out that rounding prematurely leads to accumulation of errors. By the time you're done you might have no precision left.
    This is a rather rash statement !!! Can you quote some of your regular occasions when you have pointed this out (save me searching)
    Do you mean precision or accuracy?....is it better to be precise or accurate?
    'Insignificant figures never become significant figures'
     
  9. Aug 27, 2016 #8

    SammyS

    User Avatar
    Staff Emeritus
    Science Advisor
    Homework Helper
    Gold Member

    This looks to me like it would be very helpful.
     
  10. Aug 27, 2016 #9
    It just wasn't useful to the question that I was asking. I had already said in the post that you can get back to 120. using significant figures again.

    The question was more along the lines as to what the proper answer would be that maintained the precision of the original measurement. The rounding to the significance of the original measurement bothered me for that reason, because it was technically more precise than the original measurement with identical error bounds... ##\pm1## That's why I thought it best to just give the fractional answer with fractional uncertainty that covered the entire range and maybe with a caveat that either gave the original significance of the measurement or the original measurement itself.

    Practically, if I were going to use the value in the future in another calculation which would include going back to the original measurement I would just use the entire fractional value. Rounding at the end.

    In the end, if the question is just a conversion type problem, I'll keep it in decimals, rounded to the significance of the original measurement, with the best new uncertainty as possible.
     
  11. Aug 27, 2016 #10

    haruspex

    User Avatar
    Science Advisor
    Homework Helper
    Gold Member
    2016 Award

    Yes, that is the right approach - or, at least, to keep a few extra digits of precision until the end.
    You should also be aware that when intermediates are irrational you may be forced to round them, and this might still lead to less precision than you think. E.g. a careless attack on π-22/7.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?
Draft saved Draft deleted