The Most Beautiful Experiment

[Spherical Cow]

A recent article in Physics World led me to an older article that I had never seen, called "The Most Beautiful Experiment". That article is http://physicsworld.com/cws/article/print/9746 . The article was popular enough to now become a book. For the article, people voted on what they considered the most beautiful scientific experiments. Very interesting reading. I agree with most of the top 10, but was surprised to see that Eddington's measurement of the bending of starlight (observed during an eclipse) was just an "also-ran".

Good reading if you missed it-

Norm

[xilman]

Originally Posted by Spherical Cow

I agree with most of the top 10, but was surprised to see that Eddington's measurement of the bending of starlight (observed during an eclipse) was just an "also-ran".

Do you distinguish between an experiment and an observation?

There is an element of control in the former which is lacking in the latter. Very frequently, an observation is unrepeatable in practice or in principle or both. We've only one cosmic background radiation to observe, for instance, so although we can observe it many times we can't really modify the experimental conditions. Similarly, we've only seen one cometary collision with a gas giant and so have no real way of telling whether that instance was a typical example of that phenomenon even though the observation could in principle be repeated many times for a range of different comets impacting on any of four gas giants in our solar system alone.

To be fair, Eddington's observation has been repeated many times by several different technologies and using the gravitational field of several different bodies, so has perhaps now moved into the class of experiments. The Hipparcos astrometry program, for example, had to take into account not only the solar gravitational field but also that of Jupiter.


Paul

[fivemack]

I find it very difficult to describe an astrometry experiment as 'beautiful' - they're nightmares of data collection and accounting for all sorts of small confounding effects, thermal warping of the telescope tube as the temperature drops during the eclipse, the aberration of light by the Earth's orbital velocity around the Sun, the precise measurement of the centroids of fuzzy blobs on photographic plates.

The top ten, except possibly for the torsion-bar experiment which is full of vacuum jars and incredibly fiddly quartz fibres, are all effects with fairly simple instrumentation whose results you can see immediately without having to apply second-order corrections to the results of carefully averaged measurements.

[Spherical Cow]

Originally Posted by xilman

Do you distinguish between an experiment and an observation?

There is an element of control in the former which is lacking in the latter.

You're right. Eddington's observations really shouldn't be considered in the running; can't call that an experiment. They should probably drop Eratosthenes measurement of the earth's circumference from their top ten as well, or at least put an asterisk by it.

I do agree with their number 1, though: The double-slit interference of single electrons is almost chilling, and Millikan's oil drop ought to be a close second.

Norm

[cheesehead]

Originally Posted by xilman

Do you distinguish between an experiment and an observation?

I was sorta wondering about that. E.g., astronomy, my favorite science, consists largely of observations of things over which we have no control. Nevertheless, it is well-founded as scientific. Why? There must be more to its scientificness than merely that we can compare spectral lines in a laboratory to spectra of stars, for instance.

(I'm glad you asked your question, so I could figure out mine!)

There is an element of control in the former which is lacking in the latter.

Yes, but IMO the key is whether the observer can repeatably detect a difference between two situations that differ by only one variable, or as few as possible, that can be accounted for.

Some observations are in this category, and thus are as scientific as a laboratory experiment. E.g., Eratosthenes could determine that on a specific time each year (summer solstice), the shadow measurements at Aswan and Alexandria were consistently repeatable from year to year. Eratosthenes couldn't control the passage of time or apparent path of the Sun, but he could reduce the other variables to the point where he could show there was a consistent relationship between latitudes and Sun angle, and thus, assuming a spherical Earth, calculate its size.

Indeed, if the observer confines him/herself to observing and measuring, without making any change in the object of observation, leaving all manipulation of that object to others, is that not experimentally equivalent (from the observer's point-of-view in isolation) to the case in which it is not possible for _anyone_ to make any change in the object of observation, and the observer must wait for naturally-occurring changes in the course of time?

Very frequently, an observation is unrepeatable in practice or in principle or both.

... and those observations may be less likely to be able to be treated as scientifically experimentable. But what counts is whether one can measure certain variables in such a way as to isolate them from enough other variables to allow them to be treated scientifically-experimentally.

For example:

We've only one cosmic background radiation to observe, for instance, so although we can observe it many times we can't really modify the experimental conditions.

But we can observe it in many different directions, and that is within our control!

Thus, important recent discoveries about CBR concern the small variations in intensity or temperature when measurements taken in different directions are compared. The amounts and distributions of these differences can be compared to the predictions made by different models of the initial conditions, and thus we can determine whether one model more closely fits observation than another model!

So, while the situation as a whole may not be experimental, there are observations of subsets of the whole that can begin to be treated as experimental in some respects.

Similarly, we've only seen one cometary collision with a gas giant and so have no real way of telling whether that instance was a typical example of that phenomenon even though the observation could in principle be repeated many times for a range of different comets impacting on any of four gas giants in our solar system alone.

But we learned a lot anyway!

Surely you'll grant at the very least that we learned that a cometary collision could cause readily-observable effects over an area comparable to, or larger than, Earth. That means something, and we hadn't actually seen that before. (The sizes of certain craters here and there were suggestive, but we didn't actually see the collision in any of those cases.)

We can at least say that any theory that such a collision would result in no observable effects (e.g., the theory that the comet would just go *poof* and the planet would be unaffected), which some might have formerly considered reasonable, is seriously in doubt now.

We learned how long the disruptions remained detectable on Jupiter in this particular case. Whether these might last 10-30% longer or briefer if the conditions were slightly different -- we can't determine that yet. But we did get a data point we didn't have before, and this can begin to indicate whether certain models of comet/gas-giant collisions are consistent with this point, or far from it.

[Nelson]

Cheesehead is right and it would appear that throughout the entire universe multiple experiments have been set up for us since we have no control over their initiation. Furthermore they occur at a relatively safe distance so we don't get fried in the process. Our own nuclear bombs are bad enough without having a SuperNova in the back yard as well.

But learn from them we can and will. Only our arrogance will keep us from seeing the real value in them. Quite a conincidence that the moon has just the right size to give us the kind of Solar Eclipses it does and at predictable intervals.

nelson

[ewmayer]

Originally Posted by Nelson

CQuite a conincidence that the moon has just the right size to give us the kind of Solar Eclipses it does and at predictable intervals.

Perhaps, but if it weren't [and in fact it wasn't for most of earth's history - which you will surely take as "a sign"] - you and your fellow mystically inclined brethren would surely find some other "interesting coincidence" to seize upon as an example of the "signs and wonders" you seek.

Oh wait, you already did just that, with your "amazing amino acids we can eat and digest" posting in the evolution thread.

Here's another one: isn't it amazing that the watch on my wrist has - within a few seconds every day - *exactly* the same periodicity as the earth's diurnal one? Amazing, that.

Less blatantly speciously: Isn't it amazing that the moon's rotation period is *exactly* such that the same side of it is always facing the earth? It's a miracle! And don't try to pass it off as "tidal locking" - we all know that is just a way the unbelievers have to try to ignore God.

Time to hightail it over to eBay.de and see if I can snag the bid on the amazing Wienerschnitzel with the undeniable face of the Pope magically appearing in its crispy-and-light-yet-not-too-dry breading. Der panierte Papst ... not only wondrous, but oh so delicious!