Like a lot of physics departments, we offer an upper-level lab class, aimed at juniors and seniors. There are a lot of ways to approach this sort of course, but one sensible way to think about it is in terms of giving students essential skills and experiences. That is, i's a course in which they learn to do the things that no physics major should graduate without doing.
I'm sure that other disciplines do something similar, so I thought I might throw this out there as a general question:
What are the essential skills and experiences a student ought to have before graduating with a degree in your discipline?
If you're a computational theorist, for example, you might say that no student should get a degree without debugging Fortran code. An organic chemist might say that no student should graduate without taking an interpreting an NMR spectrum. A biochemist might want graduating seniors to know how to run a gel and decipher the resulting blobby pictures.
So what are the essential experiences a student in your discipline needs to have?
I come at this from Atomic, Molecular, and Optical (AMO), so my answers involve lasers and atoms. There are two things I think students need to see at least once, and not coincidentally I try to hit both of them with the lab I do in the upper-level course.
One is interferometry. Many of the most precise measurements we make in physics are made possible by the interference of light. Whatever effect you're studying, if you can find a way to make it cause a slight delay in the propagation of light, you can measure it interferometrically, to ridiculous precision. Michelson interferometers, Mach-Zehnder interferometers, Fabry-Perot interferometers-- these are essential tools in physics, and every student should do at least one precision measurement using interferometry.
The other is precision spectroscopy. This doesn't necessarily mean using a big grating spectrometer to measure line positions, though that's fine, too. But one of the very coolest things about modern AMO physics is that laser spectroscopy makes the measurement of part-per-billion shifts and effects almost effortless. If you have a halfway decent diode laser, you can measure the hyperfine splitting of rubidium to within a few tens of megahertz, considerably less than a millionth of the frequency of the laser itself. People who do this sort of thing professionally make lasers that are stable to the part-per-trillion level, or even better.
I try to hit both of these in my module of the upper-level lab by having the students first calibrate a Fabry-Perot interferometer, and then use it as a frequency reference to measure the rubidium hyperfine splitting with a scanning diode laser. If they're careful, they can get the ground-state splitting with an uncertainty of 20-ish megahertz (out of 6800), which is pretty respectable.
(This, by the way, is one of the reasons why I'm unhappy wth 30% error in labs...)
If I could figure out a way to incorporate op-amp based feedback circuits into the lab (having them lock a laser to a frequency reference, say), I'd do that, too. Nobody should be able to market themselves as an experimental physicist without making at least one lock circuit from scratch. I can't quite work that in, though, so I stick with those two.
So, what do you regard as essential skills and experiences for majors in your field?
One of the cool, low-tech, things we did in my upper level physics lab course was to build a barometer from a jug, a stopper, and some tubing, and use it to measure the height of a nearby mountain. For comparison to another method of measuring the same thing (optical surveying in my case, since I had some experience). But it taught me that it's fairly easy to set up an experiment to measure something; you may not need fancy (or expensive) equipment.
The kids exit some $60-100K in debt - interest not being tax-deductible and debt not being discharged by bankruptcy - with no hope of employment. An organiker makes drugs, poli sci makes pimps, Fine Arts make whores, Liberal Arts make Welfare recipients. The street finds its own uses for things.
CNN says a million Mexican illegals flooded back across California's southern border to a better economic climate. Physicists could be doing jobs Mexicans won't do.
Actually a computational theorist could live QUITE happily without ever debugging FORTRAN code... wait, do you mean a computational PHYSICS theorist? or a theorist of computation?
Computational physics theorist.
Physicists are pretty much the only people in the world still using Fortran, aren't they?
Well, *I'VE* only used FORTRAN in the context of biologists, but I think some engineers (aerospace, for example) still use it...
We've got labs in most of our upper-level courses, plus a summer field course, so students have a lot of experience with hands-on work.
- Make a geologic map based on data that they collected themselves. (Includes locating themselves on a base map, measuring the orientation of rock layers, and figuring out what's going on when the rocks are buried beneath vegetation.)
- Draw a cross-section (showing what's hidden underground based on what's exposed at the surface).
- Identify rocks and minerals - with the naked eye, with a handlens, with a polarizing microscope, and if the equipment is available, with x-ray diffraction. And don't just identify them - be able to explain how they probably formed, and put together a story based on them.
Maybe this isn't what you're looking for, but I think the most important thing that undergrads get out of advanced physics labs is the ability to setup, debug, and optimize lab equipment. Without those skills, they're basically worthless with any equipment down the line that isn't turn-key. It's embarrassing to run into graduate student experimentalists that can't figure out whether an experiment is working or isn't, and WHY. There's no lab manual at the higher levels, just intuition and smarts, i.e. experience. They've got to learn early that things often don't just work when you turn them on, and a combo of deductive reasoning, tracing cables, and knob turning can often get you back on the right track.
Take your example of the measurement with 30% error. I'd much rather have a kid that could understand where that error is coming from, and tune up the experiment to make it 3% error, than a kid that can push the button on the 'canned' experiment.
Anybody who has to deal with data in which discrete events are counted must understand Poisson statistics. There are many areas where this situation arises: observational astronomy from infrared through gamma rays, space plasma physics, nuclear physics, particle physics, etc. There are times when 30% error is all you can hope for, because you simply don't have enough counts in your detector to do better (particularly in astrophysics and space plasma physics, where you don't always get to repeat the experiment).
I would also say that some kind of computer programming ability is essential. If you are a theorist, you will probably have to implement some kind of computer code to turn your theory into a model. If you analyze data, chances are there is home-brewed code somewhere in your data analysis chain, and if it isn't already there you will probably have to write it yourself, or modify the existing code so that it does exactly what you need to do. This doesn't necessarily mean Fortran skillz (although there is a lot of legacy Fortran code out there); for example, I consider C to be my native programming language (it handles command line arguments well, which comes in handy for setting up batch jobs), and other languages might work depending on your situation.
Coming from a physics/materials science background, the number one thing is probably hands on experience with vacuum pumps. At least the roughing/turbo combination, so you can operate one with out getting oil in the turbo, and clean it out when you do. Some electrical characterization is also mandatory, up to at least Hall effect. And a spin around an SEM or AFM wouldn't hurt, but that's a little scarier to have an undergrad lab loose on your equipment.
As a mathematician (specifically an analyst), I'd say know how to push around your epsilons and deltas.
Speaking as someone who has gone through multiple career changes over 45 years, the only undergraduate course that has direct application in every job I've ever held is Statistical Analysis. Students need to learn how to choose and perform the right statistical analysis techniques for any given project, and they also need to know enough statistics to properly evaluate the papers/reports of other scientists.
That said, another essential skill that was learned as part of several courses is the ability to perform a complete and detailed literature search in the appropriate discipline. At my particular SLAC, every department had weekly 3-hour evening "major meetings" for declared Junior and Senior majors. These entailed required reading and writing in addition to regular coursework. In my major, biology, all juniors were required to read and discuss certain major seminal works in the field, from Darwin to the then-very-recent discovery of DNA. We also were required to choose a fairly narrow topic of recent research, do a complete literature search on it, and then write the equivalent of a review paper on that topic. Topic selection and literature search were done in the fall semester and writing the review along with an oral report in the spring.
Every practical skill listed so far is valuable. Nor should the student be blinkered and ignore other fields (i.e. Astronomy, Biology, Chemistry, Geology, hence I'll mention Darwin below). I'll suggest a few more, and then some of the beliefs that a Physics student should have acquired by some nonlinear combination of reading, lectures, homework, exams, and labs.
(a) Read a published Physics paper in a journal and explain at least some of what it means; for that matter, read at least one article each from Science, Nature, Scientific American, American Scientist, Physics Today, New Scientist, Science News, Analog Science Fiction & Science Fact. For that matter, read and comment on a Science Blog.
(b) Keep a lab notebook, and show an ability to document and reflect on the documentation;
(c) Be surprised by something that happens in the lab; this includes the essential making of a mistake of technique or interpretation, understanding that when shown by a team member or the teacher; and going through the process again until getting it right;
(c) Get a sense of what The Scientific Method is, with actual hypotheses being tested in the lab; be willing to see how this works beyond Physics;
(d) Collaborate in a group in the lab, as Science was much more about teams that lone Mad Scientists or Victorian gentlemen of leisure in the 20th century, and wikiscience of virtual teams in the 21st Century.
(e) Present the results of the group work; this is foreshadowing of the thrill of science conferences;
(f) Know something of some important scientist, beyond the textbook level, by having done a biographical homework assignment or presentation, i.e. why did Newton probably die a virgin but Einstein certainly did not? What did Brahe and Kepler accomplish as team leaders? Who won a Nobel Prize in Physics, for what, and what challenges did he or she have to overcome?
Beliefs that should be acquired, and given a framework allowing them to be explained to others, and for pseudoscience to be tested against.
(g) Atoms are real, and are not correctly described by
classical physics (Bohr atoms would collapse quickly due to emitted electromagnetic radiation; more subtly, they would collapse by emitted gravitational radiation);
(h) Quantum Mechanics makes many correct predictions;
(i) General Relativity makes many correct predictions;
(j) Quantum Mechanics and General Relativity as they stand
now are utterly non-unified, due in part to foundational
disagreements about space and about time;
(k) Evolution is both a fact and a theory. in 2009 many will celebrate and many will deprecate Darwin's
bicentennial (of birth) and sesquicentennial (of publication of "The Origin of Species." Darwin emphasized that he made 2 major contributions, one being the FACT of
Evolution as an observed phenomenon (i.e. Darwin's
finches in the Galapagos islands), the other, quite
different, being the THEORY of Evolution by Natural
Selection (which he shared with Wallace in priority, by
gentlemanly agreement). AND as Physics students to be able to explain to a family member or stranger how Cosmic Evolution and Stellar Evolution are very very different from Biological Evolution.
(l) We have learned a great deal in the 150 years since
"Origin of Species" in ways that Darwin could not
have imagined. The Neodarwinian synthesis is a paradigm
that has now, in the flood of genomic and proteomic data,
broken down. The very word "gene" has no useful
definition any more (I say as someone recently teaching the
California State Standards on Evolution to impoverished
teenagers, using the pathetic textbook augmented by selected articles from Science, New Scientist, BBC, and the like).
(m) The universe is real. We do not live in a dream. We
are not the characters in "The Matrix" movies.
(n) The universe is lawful. In an unlawful universe, one
might as well perform endless religious rituals or wave
magic wands. That is, we don't live in the universe of Lord of the Rings, nor Harry Potter, either.
(o) The real universe, for deeply mysterious reasons, in
some ways obey mathematical laws.
(p) It is worth while for a civilized person to make an
effort to understand, on the one hand, what has been
observed about the actual universe, and to participate in
making observations within the Scientific method; and, on the other hand, to learn as much Mathematics is necessary to describe and predict phenomena that one observes, or reads about from the observations of others.
(q) More, or different Mathematics than that, is not inherently useful in normal science, but has aesthetic and
Rather than ad hominem arguments, and flaming, and trolling, I'd prefer the students to appreciate and engage in rational discussion of these (oversimplified) statements of my belief system as a scientist.
As an atmospheric spectroscopist, specializing in FTS, I can only second your recommendation of interferometery. As a grad student running a Michelson interferometer, I often reflected that an awful lot of branches of physics had to go right before I got any data.
I'd break it down farther. I assume the purpose of the course is to make the student useful in a real lab doing real experiments. With what techniques should the student develop a familiarity? The physics student should do experiments which require, in no particular order:
o Fairly high vacuum
o Cryogenic temperatures
o Phase-sensitive detection
o High voltage
o Low-noise electronics
o precision (~1 visible wavelength) measurement/positioning
o Real-time control
o Statistical analysis of large amounts of data
o Fast real-time data acquisition
o Slow, tedious manual data acquisition (lower-level labs may have taken care of this one!)
Not all at once, of course.
I'm sure I missed some. Once we have a complete list of techniques to learn, we can come up with a set of experiments that teach them.
Chiming in here a little late (I don't often find much time between classes and homework to comment on web journals) but, as my father has pounded into me over the years, the most important thing to learn in any science or engineering field is this:
Be able to recognize when you are desperately trying to convince yourself that you are correct, when you may not be.
I worked at a certain lab, which shall not be named, at a certain very important school, which shall also remain nameless, which published in a very influential journal a paper which... well... wasn't quite correct. For two summers, they had me programming simulations to try and fit to the data they were handing me, which they seemed to be cherry picking because "it looks pretty close" and damnit, it did. But it wasn't ALWAYS that close, and I, and many others, expressed significant doubts. Now, some ten years after initially publishing about it, it's slowly (and quietly) being pushed under a rug.
Sometimes great ideas aren't the correct idea for a problem, and more than anything else it is important to be able to step back and reassess the situation. Good science demands it.
Just my 2 cents.