I thought this was a rather good explanation, found on the web at
http://www.broadcasttechnical.com/articles/23mysteriesoftheshieldedloop.htm:
"MYSTERIES OF THE SHIELDED LOOP REVEALED!
There’s a variety of antennas that you can use for AM reception at the studio: many engineers have used a whip antenna, usually mounted on a ground plane. I’ve seen automobile antennas used in this way. Some favour a longwire antenna, but I’ve always preferred the shielded loop. It’s easy and inexpensive to make one, and although they’re not particularly sensitive, their unique noise- and interference-cancelling properties mean they can give surprisingly good performance in difficult situations. How come? As a matter of fact, once you start to look more closely, many people start to wonder how come they work at all! This month I’ll try to explain their secrets.
The first question many have runs thus: if the doggone antenna is shielded, how does it pick up a signal at all? The answer is surprisingly simple: our desired RF signal consists of electromagnetic waves, which have an electric and a magnetic component. We shield the electrostatic component only—and pick up the magnetic wave. Any grounded conductor can be used as a shield against the electric wave—if we had wished to shield the magnetic component, we’d have to use a magnetic material, such as iron, steel, nickel or even mu-metal. And sure enough, if we use a piece of steel electrical conduit for our shield, we won’t get much of a signal. Copper, on the other hand, makes an excellent electrostatic shield, without affecting the magnetic field, so that’s what we’ll use today.
One aspect of that shield that's bound to confuse is that there must be a break in the loop--
otherwise the windings inside will effectively couple to a shorted turn, and you'll get little or no signal coming out. The shield must be connected to ground or it will be effectively invisible, and will provide no shielding action at all. Depending on the details of construction, it may be desirable to switch the ground connection to the shield on and off, allowing the antenna to serve as a shielded or unshielded loop.
Since, in its shielded form, the loop is picking up only half of the electromagnetic wave, that explains why its sensitivity is a bit low. The surprise is that the received noise is usually attenuated even more, and that’s because most electrical noise is electrostatic in nature. An added bonus is that the rejection nodes of a well-constructed shielded loop are very deep—perhaps –25dB! (Incidentally, this explains why the shielded loop is so often used in radio direction finders.) Often, when we’re faced with a situation involving nighttime interference, we can benefit by forgetting about peaking the desired signal, and instead concentrating on nulling out the interfering ones.
If you need more sensitivity, you can resonate the loop by experimentally applying a small tuning capacitor--no more than 500 pF or so--in series with the loop. You’ll know when you reach the right value—the output level peaks up quite sharply. One precaution with this arrangement, though: it is quite easy to achieve a loaded Q high enough to lop off the sidebands, which will result in a loss of high-frequency modulation content, and distortion there too.
Received signal strength is more or less in proportion to size: twice the size, twice the signal. The optimum number of turns to use is counterintuitive—more turns does not equal more signal. As a matter of fact, signal strength drops off pretty quickly past the optimum number. This is because we're typically trying to match into a receiver front end that has a fairly low impedance--say 50 to 100 ohms. More than a handful of turns results in a high impedance device, and leakage capacitance to the shield starts to become significant, too. Flatter results across the broadcast band can be achieved by using three turns or so in the body of the loop, and connecting a balun—a balanced-to-unbalanced transformer—at the output of the antenna. This improves the impedance match and balance, because if you look at it carefully, you’ll see that the loop itself is essentially a balanced circuit. By inserting the balun, you’re providing the right type of balanced load for this antenna. Ten turns or so on the ferrite toroid of your choice, bifilar-wound, makes a very nice, compact, self-shielding balun.
So there you have it: the shielded loop, unplugged!
