One of the most important skills a physicist can have is to estimate the size of various effects without doing a detailed calculation. This basic skill is one of the hardest ones to learn, mostly because it is very easy to disbelieve: after all, you didn’t do a real calculation. But at the same time, it is one of the most important skills one can have. These estimates will tell you if you are going to be wasting your time doing a calculation or not. On the other hand, they migt give you an enormous amount of information without having a fundamental system of equations to solve: let us say, you don’t know if there is a relation between some set of phenomena, but order of magnitude estimates migt be able to do that for you. I will now describe various instances of these order of magnitude estimates to show you how powerful they are.
The first thing one learns in physics is that units are important. Physical quantities are not dimensionless numbers, they are measured and compared against other similar quantities, and if you decide that you have a standard object to compare it to, you call that a unit. For example, the meter was originally defined in terms of the circumference of the earth. But apart from the fact that we live here on the earth, the earth is just one more object that one could compare lengths to. In the old days, one could compare lengths to the length of the arm of the king or queen. Every time one had a new king or queen, one had to make a new set of units. I’m sure the merchants were not happy with this, which is why comparing to the earths circumference is a better option. At least it is more universal.
By the same token, a second derives from dividing a solar day into 86400 units of measurement, each of them called a second. The number 86400 is made of twenty four hours, each of which is divided into 60 minutes, each of which is divided into 60 seconds, and a second is not too different from the time that it takes you to say ‘one second’. Why did we choose to divide the day into 86400 pieces rather than 100000? After all, the second number is much nicer to do arithmetic with. However, you can instead blame the babylonians, because their numbering system was in base 60! This sounds bad by modern days, but 60 is exactly divisible by two, three, four, five and six, which makes divisions into equal parts with small integer denominators a lot easier. I’m sure babylonian accountants were very happy with that.
Finally, there is the unit of mass. In the MKS system, it is given in Kilograms, and a kilogram is the mass of one liter of water. A liter is a unit of volume. Why did we choose water, and not gold? Because most people have readily access to water and because water is liquid at room temperatures. That way the measurement of volumes by matter is more precise. That is why.
So all of these units are comparisons to randomly selected objects that we encounter every day and we consider important. Perhaps one of the most interesting statements one can make about physics is that after you have units for mass, time and length, the rest are derived concepts. That does not mean that other units are not useful. But it does mean that one can do an absolute comparison of properties to some controlled experiment.
As I discussed previously, once you find that the speed of light is the same for everyone, your units of time and length are related. Therefore, you can choose units where the speed of light is one. In this system of units, you measure lengths in light-seconds or light-years.
Similarly, once you discover quantum mechanics (I promise to talk about this in the future), you can see that energies can be related to inverse times. The constant responsible for that relation is Planck’s constant. Another pretty universal measurement.
In particle physics, it is customary to set
Where the h with the bar through it is Planck’s reduced constant: h divided by 2π.
A system of units with that property is called a system of natural units. The usual unit to compare to is GeV, a unit of energy whose numerical value is very close to the rest energy of a proton, or eV : the kinetic energy that one electron gets when moving through a difference of potential of one volt.
So now, let us talk about estimates. How do we put information together and make predictions?
Let us say that you have a gas, with some density ρ. The density is measured in 
These are units of velocity squared! So you would imagine that for the gas, the velocity unit given by
is rather important. But we know of some other velocities that are important in gases: the velocity of sound for example. If you don’t have anything else to go on, you could not be blamed for believing that the estimate v above might have something to do with the speed of sound. Indeed, when you estimate it, it differs from the true speed of sound by factors of order one (sometimes ). However, it is not exact. And the speed of sound is not just a multiple of v with a universal constant for all materials either. So you might imagine that the ratio of v to the velocity of sound might be important and that it might tell you something very important about the gas. Indeed, Laplace worked this out, and he found that the two numbers were related by the thermal properties of the gas in question. Indeed, one can use this equation to show that hot air transmits sound faster than cold air.
You might make similar estimates for a length of string under some tension. Indeed, that is how one tunes guitars: one varies the speed of waves on the string by modifying the tension.
Another favorite of mine is the following. Assume that someone gives you a cross section for absorption of light or some other particle by some dust, this is usually called σ. This cross section is measured in units of area. Let us also assume that you have some density of scatterers, this is measured in units of inverse volume and the usual symbol is n. If you take the cross section and divide by the density, you get a unit of length.
This unit of length is an estimate for the penetration thickness of the dust, or the mean free path before you have a collision. So if your dust or material is substantially longer than λ, you will absorb most particles. The penetration length of light in milk is rather small. This is why a glass of milk looks opaque.
Another fun example one can consider is the lifetime of the Z boson. I picked this one because this week I have been doing this calculation in class. The Z boson couples to fermions (matter), with a dimensionless coupling constant g, which is not too different from one tenth. If one is to estimate the decay rate of the Z boson, one would guess that it is estimated to be 
Now, these estimates are important for various day to day problems. Let us say that someone tells you that they can deliver to you a home nuclear reactor, and that the failure rate is small and their lifetime is long: they don’t need work on them for 10-15 years and that after that you have to call the technicians to refresh the fuel. Now you have to imagine that there is a one in a million chance of someone who decides that they can fix their home nuclear reactor themselves by hitting it with a hammer. Thereby saving a pretty penny from not calling the technicians and paying for them. Would you grant the company that makes these reactors a permit to sell them?
Shores of the Dirac Sea 
is ‘MKS’ really an official abbreviation for anything, or should we just call it the SI-system and nothing else?
Dear Anders:
MKS is an acronym for Meter-Kilogram-Second: it is supposed to remind you what the basic units are. The other common SI system cgs is centimeter, gram, second.
Dear David,
it’s a nice article but I agree with anders you that you are confused about the terminology of units. SI has always been the same thing based on MKS units while CGS was always the other system.
The SI-CGS system is an oxymoron. See
http://en.wikipedia.org/wiki/International_System_of_Units
So MKS is an old-fashioned way of calling SI but CGS is always the old system based on CGS.
Best
Lubos
The meter was either a ridiculously small fraction of an impossibly exact value for the Earth’s circumference or the length of a believable one-second pendulum – your choice.
The next generation of atomic clocks is sensitive enough to be skewed by gravitational divergences of less than a centimeter altitude – the moon orbiting overhead, lawn sprinklers, a fat lab technician. Time will not be synchronized through space.
Six of seven SI quantities are reproducible from first principles. Mass has no fundamental link to natural constants, Newton’s G cannot be calculated, the standard kilogram is a physical object whose fiducial value drifts over time. The Standard Model arrives massless. The obvious conclusion is that mass is an artifact of observation that should be normalized out of physics.
Your reactor scenario is not frivolous. The H*Y*D*R*O*G*E*N car will have housewives making 5000 psi or deep cryogenic connections. Hydrogen has the largest LEL-UEL range. Hydrogen is the ultimate leak through defects and by diffusion. Girls with Guns or housewives with hydrogen – don’t do it.
Hi Lubos:
My knowledge on exactly what the SI system encompasses is quite old: my high school books were probably outdated, not to mention the books that I inherited from my father (who is an engineer). I agree with you on this one. Old habits are hard to leave behind, unless there is a very good reason.
I’m so used to changing between units, that for me they are more or less equivalent. I grew up in meters, but now I’m measured in feet and inches. I used to weigh myself in Kilos, but here I have to do it in pounds. If I need to describe distances, I would prefer meter and Kilometer, but instead end up talking about miles and yards. I did grow up drinking litters of milk, and filling gas tanks with gallons of gasoline. I also grew up with Pressure measured in milimeters of mercury, rather than KPa.
Dear Uncle Al:
The home nuclear reactor is not frivolous. This was part of a discussion we had here a couple of weeks ago about alternative energy deployment and removing the cost of having an electrical grid. Such a device could give you electricity for 15 years with no maintenance, however, one in a million events can have catastrophic consequences for neighbors, especially because one has to estimate levels of ignorance. One then needs to do an actuarial calculation of estimated risk of deploying such technology. For these calculations one never gets an exact answer, but an order of magnitude estimate is quite achievable and it is the best one can hope for.
Dear David, well, that’s a part of culture and your cosmopolitanism. I’ve read some important CGS-based books and textbooks, too. They have a very special mysterious archaic flavor to them – that’s how Gauss was doing things etc.
Pounds and inches are a completely different story, and Herr Fahrenheit has paradoxically made it in the U.S. only. There are currently 3 countries in the world that haven’t adopted the SI units yet: Liberia, Myanmar :-), and the United States of America.
These are irrelevant small perturbations of the culture, anyway. I hope that people will eventually adopt the Planck units or string units, whichever will look better when it is decided how the extra dimensions are compactified.
Nice post. Perhaps the term “dimensions” rather than “units” is better. Units and particularly unit conversions are annoyances. Going by my kids’ reaction to HS physics, mindless study of units possibly accounts for the general aversion to physics.
Dear Morris:
There are two different issues here: dimensionful quantities refer to some property of a system that can only be measured by comparing to something else with the same type of dimensionful quantity. Hence, dimension can be length, volume, current, etc.
Units are conventions on what to compare to. Annoying as it might be, units are important for engineering purposes. This second point is important, because if you want to communicate size precisely, you need a common base.
There is this one wonderful piece of comedy in the movie Spinal Tap, when someone accidentally writes inches instead of feet for a stage show prop, and they get exactly what they ordered: a 24″ replica of Stonehenge.
There is a sad episode of an engineering spec being written in inches, when NASA was expecting it to be written in cm. The shield that was needed was not thick enough, and the spacecraft was destroyed on mars landing.
Dear Mr. Berenstein
As a retired engineer I am well aware the utility of units. But dimensions are fundamental to understanding whereas units can be looked up and a sence of scale picked up by osmosis. Dimensional analysis is something you probably take for granted but something that is very useful for me in areas of physics which are new to me e.g. scattering X-section.
Perhaps you will consider posting someting on the derivation of Planck scale units; I can make the connection for dimensions for length and mass from Compton wavelength and momentum but how does G come in?
Thank you.
Dear Morris:
I was being particularly pedantic about the distinction between dimensional analysis and unit systems. Dimensional analysis is something that we try to teach undergraduate students in their first course in physics, and I think they finally start getting it when they are in graduate school. The only sense in which for me it’s natural is that I have been immersed in them for over 20 years. In some sense, manipulation of units can do the same job.
In the post above I also cheated: I was using known facts of physics to pick at problems and guess the right answer. If a problem is about classical gases, the speed of light will not matter when describing sound. Otherwise any number would have done. Similarly, classical gases would not immediately trigger using Planck’s constant.
I’ll talk about Planck units in a future post.