Crystal radio antennas have some special requirements:
           a.  large frequency range
           b.  large signal gathering capability
           c.  efficient in getting and then transferring the received signal to the set.

     Your basic problem is normally just getting enough wire in the air to have a resonant circuit.  For 99 percent of us, the answer is:  you can't.  Even the shortest resonant antenna for the top of the AM broadcast band, a quarter wavelength Marconi antenna, would have to be about 137 feet long;  the bottom of band calls for about 425 feet.    Unless  you are one of the fortunate few to be able to erect a bona fide antenna farm  everything you do from this point on will be a compromise of one sort of another.  This means you have to start thinking about your antenna system rather than just what you hang up outside the house.

    First, the antenna itself:  I recommend you use stranded copper wire, at least 20 gage.  Try to get at least 50 feet of it "out the window" and up in the air as high and as far from the house as possible.  My experience is that anything shorter gets real hard to tune, and gives away a whole bunch of signal to boot.  Longer is better, higher is better.  Use plastic strips or ceramic separators or whatever you can get to ensure the antenna wire is not touching anything that could be grounded, including the tree you may be using for a support.  I usually just use stranded wire with a tough plastic coating, wrapping it over tree limbs as I go, and connect it at the far end with heavy nylon fishing line.  Do not go over or under power lines.  If you are coming in through the bottom of a window, as many  of us do,  be sure to use insulated wire for the final lead in.  If you have to put it inside, such as up in the attic, make sure you are not hanging it under a metal roof or ridge vent, otherwise it could very well shield your wire from the signals you are trying to get.  High straight antennas are probably best, but do the best you can, and don't be afraid to bend it in order to get more wire out there.  Contrary to popular wisdom, you can tune a bent antenna. If you like to try out different stuff, put an alligator clip on the end in the shack to make quick connections to your latest xtal wonder rig.  If you replace your antenna, or otherwise have room, a second antenna can come in handy, if for nothing more than comparison purposes.  If your main antenna is mostly long and horizontal, a shorter, and more vertical antenna might give some pleasant results, particularly with distant stations and higher frequencies.  Since you will probably be using what is considered a random wire antenna, connected at one end to the set, it may be of some value to know that the horizontal part of the antenna "points" in the direction of best reception off the free end.

    Next, get a good ground.  This is the other half of your antenna, and a good earth ground is essential.   By earth ground I mean a good electrical connection to the earth. (I once had a student who tried just laying the ground wire on the ground - didn't work) I know of several hams who invest more time, effort and money in their ground than any other part of their antenna system.  Mine is currently a tight connection to a close by cold water pipe.  If you have plastic pipes, this doesn't work.   I have also used pipes hammered into the ground, the ground rod the telephone company provides where their service comes into the house, and even the screw holding on the cover of electrical plug cover plates in a pinch (this is usually electrical ground, and may even lead electrically to the earth somewhere - works better for some locations than others).  I haven't resorted to burying assorted lengths of wire in the ground yet, but many dedicated types do.  If you are located in the desert or on a granite mountain, and good earth grounds are hard to come by, you need to read up on dipole antennas, ground radial systems and counterpoises;  good luck.  If you can not make a short connection to a good ground, and your ground lead is long enough to act as an antenna itself, you might want to try this trick:  use a piece of coaxial cable between the set and the ground connection, using the inner core for the connection; connect the shield to the core at both ends with a capacitor, say, about 0.01 uF or so, to shield the ground lead from rf.  I haven't tried this myself, but it makes sense, I think.

    Whatever you have done up to this point, your antenna is (1) too long or too short (usually the latter),  (2) is not resonant at the frequencies of interest, and (3) will not efficiently transfer whatever signals you get to your set.  Guaranteed.  Despite all these deficiencies, it will probably work pretty well, at least on the stronger stations.  If you are happy at this point, declare victory and quit.  If, however, you have a free hand left with which to flog yourself while you tune with the other,  or really want to go after them, read on.

    The first quick fix you can make is to effectively add additional length to your antenna by putting a large coil of wire between the antenna and the set.  Wind 50 to 100 feet of 20 to 24 gage wire around a core between 2 and 5 inches in diameter, and tap it about every 10 turns or so.  Attach one end of this coil to the antenna.  Use an alligator clip to connect the xtal set to various taps until you get the loudest signal.  While this doesn't add much to the antenna in the way of length, it can make the antenna closer to resonance with the frequency you are trying to dig out.

    Quick fix number two, which can be done after or in lieu of number one, is to add some capacitance between the antenna and the set.  This seems to work best if your antenna is too long for the frequency of interest.  Radio Shack  uses this technique in some of their Bunch-in-One electronic project kits, using a fixed 10 pF capacitor for a short antenna and a 100 pF capacitor for long antennas.  Short refers to antennas 50 feet or less in length.  I prefer to use a variable capacitor if I go this route.  Putting the capacitor in series with the antenna this way lowers the inherent capacitance of the antenna.  One of the more popular  designs a number of years back put the variable capacitor in between the xtal set and the ground connection.  If it is the same size as the main tuning capacitor in the rig, you can even gang them and make tuning a bit less tiresome, with little or no adverse effect on results.  This actually works very well.  If you are using a set that has the detector or the antenna attached directly to the top of the tank circuit, and don't want to make  taps in the tank coil for better impedance matching, this fix might be(I should say is) essential, and is certainly worth trying.  The problem with connecting the antenna to the top of the tank circuit coil is complex, to say the least.  I observed that the first effect is to lower the resonant frequency of the tank circuit drastically, even with a very short (a few feet) antenna.  If you're going to try to get away with this, and don't have a variable capacitor to spare, you can make a gimmick capacitor by twisting a piece of wire connected to the set around the bitter end of the antenna.  I did it another way, using some 18 gage twin lead wire; turned out that the capacitance of the twin lead was about 1.2 pF per inch, and the circuit resonance from the antenna was not affected if you used enough twin lead (one side connected to the set, the other to the antenna) to give you about 1 pF per foot of antenna.

    Quick fix number three, which is an excellent idea even if you don't do the first two, is to connect the antenna to a tap on the tuning coil.  This gives a better match between the impedance of the antenna and the tuned circuit of the rig.  Techtronics does this with their nice little crystal set, providing two places to connect the antenna (really three, but they don't mention the third tap, for some reason.  It is between the other two.)  Probably the absolute worst place, in my opinion,  to have the antenna connected is at the top of the set tuning coil along with  the connection to the detector, yet that is what several web page offerings would have you do.  The tap you use will have to be determined experimentally, but I find that low on the coil close to ground helps.

    Quick fix number four, which I recommend with fix number one:  Use an antenna coupling coil .   Wind  the coupling coil on approximately the same diameter core as the main tuning coil, using  about one turn for every six in the main tuning coil, and then put it in line (on the same axis) with the main tuning coil, with the ability to move them apart.  You will get better  selectivity, and can sharpen it up a bit more by moving the two coils apart - what the trade calls  loose coupling.  Oh yeah, the antenna and ground are connected only to the coupling coil, although you might try connecting the ground to the set as well.  The "one for six" rule is just a rule of thumb.  If you don't mind experimenting a bit, pick the number of turns on the coupling coil that gives you the best coverage across the band, particularly if you are going to use your set without a tuner.  It will still be a compromise, and the best number of turns will depend on the length of your antenna and your location, but will give better results.  Of course, you can dispense with the coupling coil if you are inductively coupling the antenna coil directly to the set's tank coil as in the basic tuner below.

    If you have tried one or more of the tricks  above, and are happy, fine.  If not, I don't blame you, so it is time to get serious about tuning the antenna and coupling it to the set for optimum sensitivity and selectivity.  In other words, it is time to talk antenna tuning and coupling methods, subtitled "how I went from 5 to 50 stations , and then to 98, heard on my xtal set".

    A properly employed antenna tuner will do a lot of "heavy lifting" for you.  First, it can make your antenna resonant at the frequency of interest, and bring up the signal level at the same time.  Second,  it will screen out a lot of unwanted stations, including those out of band ghosts.  Third, it will transfer the signal efficiently to the main tuning and tank circuit, and do so such that you can vary the selectivity of your set as a bonus.  This seems like a set of tall orders, but it isn't nearly as hard as it may seem.  I will discuss mostly sets using air core solenoid tuning coils (that is, toilet paper rolls wrapped with a single layer of wire or the like), since that's what I usually use and have the most experience with.  Before we start, let me say that an antenna tuner is not just another stage of passband filtering, such as the Miller 595 crystal set used.  A key to a passband tuner design is when you see a two or more section variable capacitor, with one section in the gang working with one coil and the other section working with the second coil.  Also let me add that an antenna tuner is essential for serious work.

    Basic tuner:  Wind a coil on  a core the same diameter as your xtal set's main tuning coil.  You may want to add about 30 to 50% more turns, but it should be at least as many turns as are on the main tuning coil.  Put a tap at least every 10 turns (every 5 is better).  Put the two coil axes in line, and connect the end of the antenna coil that is closest to the xtal set to ground.  Attach the antenna to taps on the antenna coil using an alligator clip until you get the best signal.  No direct connection to the xtal set is needed.  You should find that you can separate the two coils by a distance of one coil diameter or more and get a good signal; in fact, moving them closer together may even make things worse.  The closer the antenna is tapped to ground on the coil, the higher the frequency.   You can leave the unused antenna coil turns "dangling", as it were, or attach the antenna to the other end and short the unused turns with the tap - both ways are acceptable.  You should be able to notice a difference in signal strength of a station by moving just one or two taps along the coil.  If you find that you never need to use more than half the coil for the best signal, even at the bottom of the AM band, don't worry;  you may need them later.  In this setup, you are essentially making the antenna appear the correct length, so you are in effect adding length to compensate for the short antenna.  If you can't inductively couple the tuner to the set's tuning coil, you can attach your earth grond to the chassis ground of the set, and attach the ground end of the antenna coil to a suitable tap on the set's tank coil.

    The Teenie Tuner:  This is a variant of the basic tuner, using a coil wrapped around a ferrite core.  Here's how to make one:  Get a ferrite core of some sort, either a rod or a bar.  Wrap some typing paper over it and glue the paper into a sleeve - put a spacer between the paper sleeve and the core while winding the coil so that the core will slide easily in and out of the coil after it is wound.  Over the paper sleeve, wind a coil of some 50 to 90 turns of magnet wire in a single layer;  no taps needed.  Glue or tape the coil onto the sleeve.  Now, mount the sleeve only on a support, such as a scrap of wood, or in line with the set's coil.  Connect one end of the tuner coil to the antenna, the other to ground, and vary the tuning by sliding the ferrite rod in and out of the coil.  You should get a range of inductance of about 10:1 this way, effectively lengthening your antenna as he rod is pushed further into the coil.  As with the basic tuner, you can set the tuning coil alongside the main coil, in line with it, or connect it's ground end to the antenna tap of the set.  Just for grins, affix the teenie tuner to the back of your am superhet and see the improvement in your reception.

    Series tuner:  Here we add a variable capacitor, the same size as the one in your xtal set, in series with the antenna coil used in the basic tuner.  Traditionally, you put it between the end of the antenna coil and the ground connection.  Now you have two tuner adjustments to make (not counting distance of the antenna coil from the main tuning coil).  For the first time, I recommend you get the set tuned to a moderately strong station, then put the tuner capacitor in the mid position, and then find a tap which allows you to hear the station.  Then you can use the capacitor to bring in the station more sharply.  You will probably find you need to use more turns of the coil than you did with the basic tuner, since the capacitor "shortens" the antenna.  Ultimate signal strength may not show much improvement, but it should help selectivity.  This setup is a favorite of the "pros".  As before, the antenna and main tuning coils are in line.  You should have to make fewer tap changes with a series tuner compared to the basic tuner, since the capacitor does some of the tuning for you.

    Parallel tuner:  Use the same coil and capacitor as with the series tuner, except this time the capacitor is connected to both ends of the coil.  One end of the coil, the one closest to the set, is connected to ground, and the antenna is connected to the coil tap that gives the best signal;  incidentally, I have seen a few circuits where the antenna just connects to the other end of the coil from the ground - I don't recommend this, even though it works fairly well, sometimes, maybe, I think.  Adjust the capacitor as needed to peak the signal.   This setup is often used when the antenna is "very" short.   My recommendation to add turns to the antenna coil will sometimes let you use series tuning across the whole band.  If your two coils are the same length, I think you will find that the series arrangement works best at the high end of the band, and the parallel arrangement is needed at the bottom; the longer your antenna, the lower in frequency you can use a series tuner.  As with the series tuner, a parallel tuner gives you more selectivity than does the basic tuner.

    A broadband tuner:  This is not a classic design, to my knowledge, and I stumbled on it by accident, but it does a lot of heavy lifting, and helped me well over the single digits in stations heard.  I thought I was copying faithfully a design out of an ARRL handbook for dealing with random length antennas, but ended up with something a bit different, and won't try to analyze how it works, but here it is:  Don't sweat the coil/capacitor specs; what you have in the series or parallel tuner will work fine.  However, the connections shown going to receiver antenna and ground are actually connected to that coupling coil I mentioned up in quick fix number four.  The tuner's orientation with respect to the set is not a factor, since the signal is transferred via the coupling coil.  It will generally be closer to the main tuning coil than described in the three tuners above.   I like this tuner arrangement for casual listening, as I don't have to play with it much over the band.  Find the settings that work best for you.  It is a particularly good setup to use when you are finding out about a new xtal set, and don't want to have to mess with the tuner every hundred kHz or so, as you do with a more selective tuner, or don't want to "tailor" a tuner just for the new rig.  This setup works well when you can't inductively couple the antenna tuner to the set.  It also works without the coupling coil by connecting the end of the antenna coil to a low tap on the main tuning coil of the set.

    While I mentioned putting the two coils on the same axis, you can also place them side by side, as Mike Tuggle does with his Lyonodyne #17 (getting 2500 mile contacts), using an antenna tuning coil wound with Litz wire on a ferrite rod;  just make sure they are parallel for maximum coupling.  You can vary the coupling by either separating them from each other or by rotating one of them from the parallel position, as is done with variocouplers.  Minimum coupling occurs for a given distance when the axes of the two coils are perpendicular.

    For you people using  toroids in your xtal set, I have only to offer what  you see in the "quick fixes".  My tuners essentially rely on loose coupling to the main coil, and I really haven't played with the other type tuning coils enough.  Al Klase uses toroids in his very fine crystal set, and you should look at his page for coupling to that.  Also look at his work on XTAL Sets 102 for ideas.

    In your quest for sensitivity and selectivity, you may now be ready to try a wave trap, sometimes known as a QRM coil, to get rid of offending, overbearing bandmaster stations.  The premise is simple; trap the strong signal and it won't bury the weaker ones.  While I can separate two local stations that are 40 kHz apart, with one of them at least 3 dB louder than the other one, I often can't eliminate the louder station entirely, and can often hear it in other parts of the dial except when I am between stations.  Then there is the signal from the secret Radio Habana transmitting site, which pokes through at odd times.  Crystal radios tend to go with the signal that is the loudest, hanging on until something better comes along.  A decent wave trap just might be the answer.   .   One basic design is around a high resistance coil, 32 gage wire or so, in parallel with a variable capacitor, and a 15 turn link around the coil which connects antenna to set (or wherever you want to try it.  The idea is that the resonant LC circuit not only is set into motion by the offending signal but also dissipates it in the high resistance windings of the coil.  The link lets the other signals bypass the trap with little loss.  Here is a link to a nice short page on wave traps.  Medium wave wave trap.   These traps, with no coupling coil, work best for out of band intruders when used as in-line traps in the antenna line, or when inductively coupled to your set's tank circuit.  I think the one I use is better for in-line use, and you can find a description, picture, and discussion of it and trapping in general here .  I realize that a wave trap might not be considered technically a part of the antenna system, but as long as it is in front of the main tuning circuit,  I will consider it so.  Wave traps can be used as either rejectors, to block an offending station,  or as acceptors, which can be sometimes used to boost a desired station or, more commonly, to add another stage of filtering to increase overall selectivity.

    Final warning!  When you do anything   to the antenna system, you can expect it to affect some other aspect of the crystal radio's tuning.  When you are ready to start making adjustments, have a comfortable chair, a tall drink, maybe something to make notes with, and the time to be patient.  And oh yeah, I didn't even mention the possibility of using different antennas, different grounds, loop antennas, no ground at all.... and did I mention lightning arrestors?

    If you noodle around on the web a bit, you will find some excellent antenna articles.  I am just trying to get you started.    As I said before, every antenna is a compromise - just try to get a setup that works for you, and then you can get back to building and testing your own crystal set creations.

    A word or two about small loop antennas, portable antennas and the like:  You will find at times references, plans and testimonials in praise of the small loop antennae.  If you live near strong stations, they may work for you.  Where I live, they are almost useless, and do require a ground.  I personally believe that if you are going after the distant stations with a loop you will be very disappointed.  These are specialty antennas and have limited value - but where they do work, they get high marks.  As an aside, I suspect that if you can't get at least 30 feet of honest antenna out the window, you might try a large diameter loop antenna instead.  If you are new to this business, start at the top with an honest antenna (please try for at least 50 feet out the window), and then, when you start getting experience, and have a set you know works, try what you like.  Trying to hook up a first crystal set with a small loop is an invitation to frustration.