Much has been printed in QST over the years on loop antennas. Experimenting with wire antennas is a favorite pastime for me. I recently had great results with a delta loop on 10 meters and a rectangular loop on 20 meters, (50+ countries in 3 weeks of casual operating!), then I decided to take the plunge and put up a 80 meter full-wave horizontal loop which would allow operation on all HF bands.
After gleaning all the info in the articles referenced below, and adding my own twist, the antenna would either work or be a "cloud-burner".

I am happy to say this simple antenna far exceeded my expectations!
What I found to be so appealing about this antenna was that it was fairly economical and easy to build and install, works on all HF bands and requires no special feed networks, only a transmatch, coax and some space!

80 METER HORIZONTAL SQUARE LOOP

Length of a full-wave 80-meter loop is about 270 feet long (1005 divided by frequency in MHz) or about 67 feet per side. I use "about" because exact numbers are not that critical according to my results.
In my opinion, when constructing antennas, not only is the old saying "the higher the better" true, but also "the longer the wire the better" may also fit some loops.
Since I lived on 10 acres in the country, I decided to make my horizontal loop longer to start with to better fit my backyard. So, my "longer" loop is about 1.25 wavelengths on 80 meters (2.5 on 40m, 5 on 20m and 10 on 10m) and is installed between 30 to 40 feet in the air.
Scaled-down versions, say 75% of a wavelength may also work fine if you don't have the room for a full-wave or longer antenna.
According to antenna experts, a circular loop is "ideal", but impractical for most hams. I found a square or even a rectangular loop is easier on the pocketbook and muscles to put up and would provide about the same results.
To support my loop made from salvaged telegraph line wire from the Yukon Territory (just think about the stories this wire has already told!),
I used my 50-foot tall tower and three masts, each 35 to 40 feet long, made from 2 inch galvanized water pipe. Each support is "supported" by one ¼ inch diameter steel guy wire attached by a u-bolt in the opposite direction of the wires "pull" and a small pulley with 3/8-inch diameter rope for hoisting up the wire to the top of masts (Figure A below).

Figure A: Detail of a Mast Support

The telegraph wire is #6 AWG copper-clad steel and not all that easy to work with, but the price was right.
For the feedpoint connection I used a 1-1/2" PVC pipe T terminating the antenna wires to a 1/4 " eye bolt as used on some commercially made baluns.

RG 213 coax (chosen for strength, durability and because I may use an amp) terminates on the eyebolt nuts with two flat washers.
The coax is taped to a ten-inch long bottom extension of the PVC T to remove strain on the hanging coax.
Silicon caulk was then applied to the connections for weatherproofing.

For antenna insulators, I used porcelain electric fence insulators.
Once the support masts are complete with guy wires and pulleys and installed, raising the wire becomes a one-man operation.
On my tower I installed a six-foot long 4x4 painted wood post hanging off near the top of tower for one of the four required supports (Figure B below).

On the post end that is further away from the tower I used an electric service entrance insulator fastened by u-bolt to "float" (ref. ARRL Antenna Book page 5-17) the antenna wire as with the other three supports. I wasn't sure if all "floaters" would actually allow the wire to float, but they did quite easily.
The wire antenna and feedline connections were made up on the ground then hoisted up each mast one-by-one with the rope and pulley. Once the wire was in the air and about a foot or two away from the masts, I merely tied off the rope to whatever was handy (i.e.- nearby barn roof, tree etc.). I only had to take up a bit more slack from one pulley (the wire pulls through all the pulleys) for final wire sag adjustments. Since my wire was very heavy duty, I could pull it tight. Your sag will depend on the type and size of wire used. Smaller gauge wires will break if pulled too tight or used on long spans - just ask me!
My loop is fed about mid-span and the coax drops 30 feet straight down into my shack.

Figure B: Detail of the Tower Support

HOW DOES IT WORK?

During the first 3 months of use, (October through December), 75 percent of my QSOs on 10 and 20 meters were either 5x7 or 5x9 reports "both ways". About 75 percent of them being with stations outside North America *(about 10% were 5x9 +20!), and about 20 percent of the total QSO's were 5x5 to 5x1 quality "both ways".
For those doing the math, call the remaining 5 percent split equally either 3x3 signals or simply "no contact at all" (you can't work them all!).
Also, my log indicates "sent" report was the same as "received" most of the time. I even broke several big pileups on the first or second call.
 
Directivity? Well, the loop seemed to work just fine equally in all directions (I'm still scratching my head!). That's what I really like about this loop!

Gain, you ask? Well, some..... depending on your choice of feedline and how high you install your antenna. L.B. Cebik W4RNL goes into a lot of detail on gain (see ref #4 below) in his article so I won't get into that here.
 
Although I have mostly tried this antenna on 10 and 20 meters, I was also pleased with a weekend of experimenting on 15 and 17 meters.
DX worked on 15 meters: KL7, HL5, JR1, KH0, RV9 and BD4.
DX worked on 17 meters was KL7 and OH1. Many Ws and VEs were also worked on 15 and 17 meters.
Both bands produced about the same results on signals mentioned above over the two-day period of tests.
I am confident this antenna will produce good results on 40 and 80 meters as well. I know it tunes 40 and 80 meters quite fast!
 
To give a better perspective on this versatile antenna, on December 29, 2001, I worked my buddy Rick KL7AK back-to-back on 15, 17, 12 and 40 meters! On 15 and 17 meters we both exchanged 59 +20 reports, on 12 and 40 meters we were up to 55 quality. Not bad for a piece of wire, eh?

I did learn however, both 300 watt manual tuners (MFJ and Vectronics) that I used took some time to tune the loop, with a couple bands requiring a lot of patience! I did not try the auto-tuner on my TS-570D since my Tucker 1.5 kW tuner easily handled the job quite fast on all bands 80-10 meters.

My tribander will remain stored in my garage as I work on a new loop design around 1200 feet long supported off of 60 foot tall power poles (but that's another article!) - when I tire of this antenna that is!
I highly recommend this antenna. Good luck with your antenna experimenting! Any and all feedback is appreciated.

Some final? words!

References:

1. "The Loop Skywire", by W0MHS, QST Nov. 1985, page 20 and ARRL Antenna Book 16th edition, pages 5-16.

2. "The Droopy Loop" by KJ7MZ, QST July 1996, page 57

3. "Loop Antennas", ARRL Antenna Book, 16th edition, pages 5-1

4. "HO-HO-HOHPLs" by W4RNL (www.cebik.com) Do a search

kl7jr@yahoo.com