Dorky Poll: SI or CGS?

I have a doctor's appointment this morning, and then class, so here's another Dorky Poll inspired by the fact that I'm teaching intro E&M:

What system of units do you prefer for E&M: SI, or CGS?

This is even dorkier than usual, so I suppose I should provide some context...

The CGS system of units uses Centimeters, Grams, and Seconds as the base usints for everything, as opposed to the metere, kilograms, and seconds of the Systeme Internationale (also called MKS in some places). This doesn't make very much difference in mechanics, but it's a big deal in electromagnetism, because the units generally associated with CGS there are very different, and change the form of the equations.

The biggest difference is the elimination of a factor of 1/4πεo from Coulomb's Law and related equations. Basically, the Coulomb constant is set equal to one, which simplifies a lot of equations.

For this reason, CGS units remain popular in a lot of E&M books.

Personally, I favor SI units, because nobody in the real world uses "statvolts" or "statcoulombs" to measure anything, and the unit conversions are a bitch. It's easier to just cope with the 1/4πεo, and directly apply the results. But then, I'm an experimental atomic physicists-- theorists and astronomers may have a different view.

More like this

SI of course. As you've pointed out the only people that like cgs are the theorists.

cgs of course, and for the record, I am a theorist. :-)

Really depends on course content. If it's supposed to be introductory and teach people basics like U=RI, then go with SI units. If it's theoretical electrodynamics & relativity, take CGS with appropriate EM units (Gauss or Heaviside). Best of all, add a section explaining how the confusion arises from the basic experiments relating EM phenomena to mechanics and what freedom you have in the choice of units.

SI. I'm a theoretician who finds having units floating around at the end of a calculation to be very helpful. Oh, look, I ended up w/ eps0 cubed--there's an obvious mistake somewhere....

By Grant Goodyear (not verified) on 02 Apr 2008 #permalink

SI.

1/4Ïε0's get thrown around, sure, but they're constants... it's not like they're integrable or differentiable functions. They're a minor annoyance I'd gladly accept in favor of familiar units.

units? units???? Dammit Chad, I'm a theorist, not a miracle worker!!!!

Nah the thing to do is to use CGS, or something that hides constants, then at the end replace e by e/4*pi*epsilon_0.

BUT it is up to the instructor to make damn sure the folks in the class get NUMBERS out at the end that are in SI. Ignoring this is not a good idea. Plus many theorists are so used to making c=1 and such, we don't even tell the students its a unit choice, and it seems like fricking magic.

And another thing. Dimensionless quantities are extremely great. Simplifies equations. But again, you need to be able to go back and forth.

When I talk to Luis, head of a research group, and I ask him what the detuning in his experiment, he gives me "oh it's half a linewidth", and that places it in context. When I ask his students they tell me "oh its some ### of MHz". Eventaully after working with them, they realize this doddering old fool doesn't know the natural linewidth of the alkali atom of the day (currently Rubidium).

But dimensionless quantities are very useful for me, you tell me you are detuned 10 linewidths, OK, then I know that there's little population in the excited state, and the effect of a laser field is mainly to produce a level shift. What is a "large" detuning? Well 5-50 linewidths often!!

Also when we take some dimensionless thing to infinity to take a limit, its cool to do the exact calculation (even numerically) and find out that often infinity is 10-100, and zero is .1-.01!

Also I tell my students we don't take limits really. Go to r= infinity? There is no field, but also you cannot get there! Give me the leading term!!!!!!

cgs...but then I never use Maxwell's equations for anything

Gosh, if I had to actually calculate stuff, I would use SI, so go with SI.

By Brad Holden (not verified) on 02 Apr 2008 #permalink

Wait, you mean there are units other than GeV and GeV^{-1}?

CGS, big time. I hate all the clutter in SI electrodynamics. Oddly, it seems that SI is more popular in intro classes for non-majors, while it seems to me as though using less constants like that would make learning the subject much more pleasant for a newcomer. There is just a certain fraction of people whose eyes glaze over when they encounter too many greek letters. CGS keeps you focused on the variables themselves.

many intro texts use

k

for the Coulomb constant, k=1/(4 pi e_0)

Thats fine till you get to problem with multiple k's,

Like a pair of ionized potassium atoms, with the bond taken as a harmonic oscillator, at some temperature T.

4 k's! Coulomb, spring constant, Boltzmann, and K for Potassium!

SI all the way. Though I am a purist, so if we could get mgs units (why kilograms instead of grams in SI) then I'd be really happy.

I also enjoy c=1, and find engineer's use of Imperial units to be quite silly. Case in point, btu per pound-mass ranken is the most ridiculous set of units I ever heard a friend use.

If you ever need real experimental numbers, SI is clearly the way to go. I've heard that when he put out a third edition of his famous textbook, Jackson switched to SI units.

It also has the pedagogical advantage of explaining why Maxwell could recognize light as an electromagnetic wave: when you derive the appropriate wave equation you get the implied velocity (epsilon_0 * mu_0)^-1/2, which happens to be c. Try to do this in Gaussian units, and that constant c has to be introduced in deus ex machina fashion.

A brief note on Perry's comment: "Dimensionless quantities are extremely great. Simplifies equations. But again, you need to be able to go back and forth." For the most part, I agree. However, it can come back to bite you in hydrodynamics. In many applications the Reynolds number (which gives the ratio of inertial to viscous forces) is large, suggesting that viscosity is negligible; however, when you discard that term you usually discard a boundary condition, and you get a nasty surprise when turbulent flows develop at Reynolds numbers of a few thousand. Dimensionless parameters are useful tools, but you have to recognize the limitations.

By Eric Lund (not verified) on 02 Apr 2008 #permalink

SI all the way. Though I am a purist, so if we could get mgs units (why kilograms instead of grams in SI) then I'd be really happy.

I also enjoy c=1, and find engineer's use of Imperial units to be quite silly. Case in point, btu per pound-mass ranken is the most ridiculous set of units I ever heard a friend use.

SI. My first instinct is to say cgs, but I think it should be introduced along the way in a special section called "college prep E+M" (to prep you for grad school). I learned freshman E+M from the berkely 2 book by the demi-god Purcell. It was in cgs and, with its emphasis on special relativity, prepared me perfectly for grad school and the tablets produced by Jackson. I learned the SI version by teaching it, and prefer it because it makes it easy to connect theory to reality. It is particularly strong when it comes to showing the magic that Maxwell discovered in his equations, as noted above in #14. But there is also the detail that you don't go to the store and ask for a 5 milli-statvolt battery! (Or a 1.5 J/C battery for that matter.) You miss some physical intuition about scale with cgs units, but you gain some theoretical insight from them.

Comment on #12. The only thing missing from your example would be to find a way to include some chemistry with a rate constant K or two.

PS - I blogged my comments about intro QM a few days ago, FYI, but I now realize that the real bee in my bonnet is about how basic thermo is introduced in all physics books. I think I will start an open thread with a few random questions stimulated by what I have been trying this week in class.

Eric - well ya can't throw away boundary conditions, thats where God is :-) (Physics is differential equations and God is the initial conditions, someone said that once).

Weirdest unit was Furlongs/Fortnight for velocity on a sophomore exam I took.......

And pray tell what the hell is a slug :-) Danged engineers......

I prefer cgs units. With those c's kicking around it reminds you that relativity might have something to do with electromagnetism.

SI. But I routinely have to deal with engineers who use units like lb-mass and kilofeet. Do you remember the recent failure of a Mars orbiter because someone apparently used the wrong measurement system?

Personally, I favor

G = h = k (Boltzmann) = c = 1

then indulge yourself

Definitely SI. Feynman points out that when definition the magnetic flux (or, "It's just H!", as Griffith likes to say), people may actually be talking about different things depending on which units they're using. CGS H and SI H are not the same...and it's a real pain for those of us trying to look at papers in both physics (where it's a crapshoot) and engineering (where it's almost all SI).

Perry (#17): Engineers in the US don't use slugs. Engineers use the avoirdupois pound (lb), which is a unit of mass in the US "customary" system, and either pound-force (lbf) or poundals for the unit of force when working in "english" units. Actually, they hide the lbf by using kips (1000 lbf) for force sort of like we use kN in SI.

By CCPhysicist (not verified) on 02 Apr 2008 #permalink

CCPhysicist- thats cool. I was an engineer first, but they had us write homework questions in one color ink, answers in another, on green paper with squares, and they took off if you made your letters wrong (i.e. like not in a drafting class).

The physics folks meanwhile were letting us play with liquid nitrogen.

Then I found out about quantum mechanics and I switched. My engineer buddies had a wake for me. They are also quite a bit wealthier too, but I have casual friday everyday of the week and get paid to do weird quantum stuff, so I'm very happy.

But literally, "some of my best friends are engineers".....

And I do love to point out to students that you can do an electrostatic measurement, and get e_0, do a magnetism measurement and get mu_0, and figure out the speed of light.

That and the fact that electricity and magnetism morph into one another when you switch inertial frames, thats easy to see and cool.

Both those things really intrigue freshman too.

Just think, written down in the 1860's, gave us all our 'lectronic gadgets, gave birth to special relativity, and by turning E and B into operators it's all cool in QM too. The spatial parts are classical really. Pretty damned cool.

Will superstrings ever get there?

In my college and grad school days back in the 60's I grew up on cgs. Never got truly into SI before I started my own business (hardly any post docs after the Vietnam War, and I was no good at selling used cars) and could delegate most numbers to others. A good friend back in the day was a partisan of the stone-furlong-fortnight system, where most physical constants were close to unity times some power of ten (as I recall). The conversion factors were impressively large powers of ten.

By triplepoint (not verified) on 03 Apr 2008 #permalink

I would have to vote for CGS (yes, I am a theorist too). The worst thing about SI is that it assigns different units to E and B.

As an experimentalist, I don't care what the theory says the number 'should' be, I know what it is, and its in volts/amps etc.
SI is much more practical.

A slug = 14.593... kg. (Google conversion)

Plank units FTW!

By Alex Deam (not verified) on 16 Apr 2008 #permalink