| Table of Contents |
A laboratory experiment is a fleeting experience. Once it has been carried out it exists outside of your personal recollections only in whatever record you kept of the data you took and the conditions under which you worked. If this record doesn't make clear just what you did, and with what results, then for anyone who wants to make any use of your results, the experiment may as well never have been done. The ability to keep a full and organized account of experimental work, as it is being done, is an essential skill in any field of basic or applied science. In the elementary physics laboratory, this translates into keeping a good lab notebook.
You won't be required to write formal lab reports on these experiments. Instead, your lab book itself is turned in for grading and comments after each lab session. Everything you've done, both in the lab period and after, should go in the notebook. It should go in as it happens. You should not think of the lab notebook as a place to do your "final writeup," but as a live logbook that you keep all through the course of the experiment preparation before lab, setup, taking data, analysis, and final conclusions. I'll have a good deal more to say on how to keep a good lab notebook in a little farther on.
You should make a habit of finishing your analysis, and drawing your conclusions, as soon as possible after you finish work on an experiment. To encourage this, your lab notebooks will be collected shortly after the lab period. They will be graded and returned in time for the next scheduled lab; the exact schedule will depend on your instructor.
Your work on each experiment will be graded on a 0 to 10 point scale. Here is how the scale is supposed to be used: a typical grade for a normal, adequate job, neither really outstanding nor really terrible, is 7 to 8 points. You shouldn't expect a higher grade unless your work is somehow exceptional or original, while a grade below 7 says that something is definitely lacking about your effort. Be sure to look at the comments your instructor makes on your work, as an indication of what you might be doing better.
Doing good lab work, and keeping a good record of it, are complex skills, and grading you on them involves the individual professional judgment of your lab instructor. Thus I can tell you only in general terms the standards against which you're being evaluated. If you don't understand or agree with the way you've been graded, take it up with your instructor but promptly; grading lab books is an inescapably subjective process, and if you ask me about your grade six weeks later, you are going to get a vague answer!
In this chapter, and the next one, I've tried to tell you some of the things that go to make up "good lab work"; as for grading, there are three roughly separate areas in which we evaluate your notebooks:
The third area above needs a little comment. No measurement is ever perfect. There's always some range of uncertainty associated with the result. In general, an intelligent assessment of the sources and amount of this uncertainty is at least as important as the result itself.
If anyone will want to use a result, certainly he or she has to know how well it has been measured. Example: you're 100 lonely miles from home, late on Sunday night, with a car that you know gets exactly 21 miles per gallon of gas; and your gauge says you have 5 gallons left. If the gauge is good to within ± 0.2 gallon, you'll start for home; but you only trust it to ± 1 gal, you may not want to. The moral is that the gauge reading is no good to you unless you have an idea of its uncertainty as well.
The whole subject of how you make reasonable estimates of the uncertainties in physical measurements, and how such uncertainties affect our evaluation of other results derived from the measurements, is called "error analysis". Note that "error" is not how far your measured result differs from some expected or "right" answer; the word for that is discrepancy. (In the real world, if you measure something, you usually don't know what the right answer is.) And "error" means something different from "mistake". Genuine blunders aren't an important part of experimental error analysis; one assumes that they can be either avoided or corrected. Error analysis, properly, is something else: it is the business of assessing what the built-in limitations of the experiment are.
Consider an example. You've measured the speed of sound in air at room temperature as 326 m/s, by observing the lengths of an air column at which resonance occurs. (You'll do this in the experiment "Resonances in Strings and Tubes" in Chapter 7.) You believe that your only significant source of error is in locating the "resonant" lengths, and that this has been done to about ± 1.5%. Then your result is 326 ± 5 m/s. You look up the book value; it is 331.3 m/s. There is a discrepancy of 331.3 - 326 = 5.3 m/s. Since this is of the same order of magnitude as your estimate of the uncertainty in your determination, you can blame the discrepancy on experimental error with a clear conscience.
But then you remember (to your horror) that the speed of sound in a
gas is proportional to the square root of the absolute temperature, so
your value (measured at 18
C)
must be reduced by 10 m/s before you can compare it with the (0C)
textbook value. Your value (corrected to 0C)
is now 316 ± 5 m/s, and the discrepancy is 15.3 m/s, substantially
larger than your estimated error.
There could be several causes for this. You might have badly misjudged how well you can measure the lengths at which resonance occurs, or you might have overlooked some other significant error source -- maybe the frequency dial on your sound source was miscalibrated. Or the difference may be a real and physically significant one, that is worth pursuing further. In any case, tracking down the cause is at least as important and interesting as what you got for the speed of sound.
Plainly the better you are at assessing errors, the better your chance of spotting and nailing down a genuine discrepancy (and more than one Nobel prize started life as a small discrepancy in an experiment someone thought he understood!). You have to avoid the temptation to quote a very large error, in order to maximize the chance that you have the "right" answer bracketed. This is every bit as wrong as it would be to quote a too-small error in order to make your technique seem impressively precise.
There's a lot more about the elementary concepts and methods of experimental error analysis in Chapter 2.