Up until the early '80's, this was a fairly easy question to answer. You would stand over the frame of a bike, and if there was an inch or two between the top of the top tube and your tender parts, that was the right size. Bikes commonly came in frame sizes two inches apart, so there was not much question whether the 21" or the 23" was the "right" size.

At that time, in the world of mass-produced bikes, the difference between different size bicycles was that the larger sizes had longer seat tubes and head tubes , so the top tube was higher. This was usually the only difference between frame sizes.

In a given model, the height of the top tube would vary, but the length of the top tube and every other part of the frame would be same, whether the bike was a 19" or a 25". A person who buys a 25" bike is likely to have a longer upper body than someone who buys a 19", so the larger rider will likely feel cramped by having the same length top tube that puts the handlebars too far away from the 19" rider. The only concession to this difference was that the better builders would supply a stem with a longer reach on larger frames, and a shorter one on smaller frames.

This all changed when the Japanese got serious about the U.S. bicycle market, and modern bikes are generally built with "proportional sized" frames. This means that the smaller sizes have shorter top tubes, and the larger sizes have longer top tubes. This is generally a great improvement, particularly for riders of "average" proportions.

A couple of other factors have made it harder to be sure of frame sizing. One is the fact that bikes come in more sizes than they used to. Where they used to come in increments of two inches, they often come in 2 or 3 centimeter increments now.

Generally, when you see a single number listed as a frame's "size" that number refers to the length of the seat tube .

A further complication is that nobody knows how to measure a bicycle's seat tube any more. Even leaving the inches/centimeters question out of things, there is the question of where the seat tube ends:

Anarchy reigns; I know of one bicycle line that made a running change in the middle of the year. You could have two bikes of the same make, model, year and nominal size, but one was 2 cm larger than the other! The only way to know was to measure them.

An additional complication is that the height of the bottom bracket varies over a considerable range, typically anywhere from 10.5" to 13"! Thus even frames that use the same system for figuring the top of the seat tube may have widely disparate stand-over heights.

When people speak of bicycle frame sizes, they generally speak in terms of the seat tube length. As mentioned above, this used to be the only variable, but with proportional sizing it no longer is. I would submit that seat tube height is no longer the most important frame dimension. More determinant of the actual way the rider will sit on the bike is the top tube length.

It is obvious why you shouldn't have a bike that is too tall to stand over with a reasonable safety margin (although even this sizing practice was not universally accepted for the first 30 or 40 years of the diamond frame.)

On the other hand, why shouldn't you ride a "too small" bike? "Because the seat and handlebars will be too low!" That was a good objection ten years ago, when tall seatposts were a rarity and quality handlebar stems were available in a variety of forward extensions but only one (short) height.

All that was before the mass production of the mountain bike. Now 250 mm and 300 mm seatposts are stock items, and a variety of excellent handlebar stems are available

There are a number of sizing systems available today, which require various measurements of the cyclists body and reccommend frame sizes on this basis. Probably the best known of these is the New England Cycling Academy FitKit. I use this system myself, but not in a blind, rote manner. The FitKit makes reccommendations for a particular seat tube length, and a range of top tube lengths with corresponding handlebar stem extensions. For instance, for a particular rider, it might suggest a 58 cm seat tube with a combined top tube and stem extension of 66 cm. This 66 cm might be from a 61 cm top tube with a 5 cm stem, or a 54 cm top tube with a 12 cm stem, or any other combination that adds up to 66 cm. Any of these combinations will give an equivalent posture on the bike. One or two combinations are particularly reccomended because the more extreme variations of stem length can cause a bike to handle strangely because of the positions of the hands relative to the steering axis.

The "by-the-book" fitting method would then be to select a suitable bike with a 58 cm seat tube, measure the top tube, and install the reccommended stem. I would submit that this approach is due to the old fashioned fixation on seat tube heights. Better, in my opinion, to find a bike with the ideal length top tube, fit the reccommended stem, and not worry about the seat tube size, within reasonable limits.

Comparing two frames mainly in terms of top-tube length is only valid if both have similar seat tube angles. Generally, each degree of difference corresponds to about a centimeter of top tube length. Thus, frame "A" with a 58 cm top tube and a 72 ° seat angle can give the same riding position as frame "B" with a 57 cm top tube and a 73 ° seat angle. This presumes that the saddle would be slid 1 cm farther forward on the seatpost of the frame "A."

The angle of the seat tube is quite important, but for any given type of bike there isn't usually as much variance as there is with top tube length, so it may not always enter prominently into a buying decision.

The fore-and-aft adjustment of the saddle permits fine tuning of the saddle position, and if there's a severe mismatch, there are special seatposts and seatpost attachments that permit a greater adjustment range.

The position of the saddle with respect to the bottom bracket is a major determinant of how comfortable you'll be on the bike.

On a bike, the weight of your body is supported at 3 locations:

Your butt and your feet are made to support weight, but your hands and wrists are not. Hand discomfort is a very common complaint among cyclists, and it is most often the result of positioning/adjustment problems.

Hand/wrist/shoulder/neck pain often result from inappropriate handlebar adjustment. It is less obvious how saddle adjustment can cause or relieve problems in this area.

Try an experiment:

This is because you cannot maintain front/rear balance while leaning forward without moving your butt back at the same time to counterbalance your upper body's forward position.

On a bicyle, much of the rider's weight should be carried by the pedals, but if your saddle is too far forward, your legs alone can't support your upper body, so you'll wind up leaning on the handlebars too hard.

Different cycling styles involve different amounts of pedal force. Racers obviously apply more force to the pedals, more of the time. The usually recommend position for a racer is the "KOPS" position, which usually works out well with seat tube angles in the 73 °-75 ° range. (Riders with long femurs would go shallower, riders with short femurs steeper.)

For recreational riders, who don't tend to pedal as hard or as much of the time, a more relaxed position, with the saddle farther back is likely to be more comfortable. As the saddle goes back, the handlebars will generally move back and up to avoid an excessively sharp bend in the torso.

Among triathletes and time trialists, there's a current vogue for unusually steep seat angles, sometimes as steep as 90 degrees. This type of frame results in the arms carrying a lot of the rider's weight, and is generally used only with æro type handlebars and arm rests.

Systems such as the FitKit also make reccommendations for how to adjust the saddle, width and drop of handlebars, shoe cleat adjustment, etc. There is one vitally important adjustment that is ignored by most of these systems, for good reason--the handlebar height. These systems cannot specify handlebar height because this is not determined so much by the dimensions of the cyclist's body, but by the cyclist's condiditioning and type of cycling activity.

Low handlebars provide a leaning-forward riding position. What supports the upper body in this leaning forward position?

Consider two extremes of cycling intensity. Contrast an old geezer who rides 20 miles a year, ambling along at 5 mph on a quiet bike path, versus a racer in a desperate final sprint to the line.

The geezer pedals no more than neccessary, coasting as much as possible, expending much less energy than a pedestrian. If such a cyclist is placed on a bike with low handlebars, the weight of the upper body will bear heavily and painfully on the wrists and hands, causing soreness and discomfort to the shoulders and neck as well. This discomfort will quite possibly lead to the abandonment of cycling altogether.

On the other hand, the sprinter's upper body is being held up by the pedaling force of the legs. Not only is no weight resting on the handlebars, the sprinter is actually pulling up on the bars to resist the downward pressure of the legs. The sprinter's upper body is perfectly comfortable with a very low relative handlebar position

Thus, the ideal handlebar height with relation to the saddle height is a function of the intensity with which the cyclist pedals. This will vary as the condition of the cyclist varies, for instance, it is often helpful to raise the bars a bit in the spring if the cyclist has been off of the bike for the winter months, then lower it back down a bit during the course of the season. The racer will usually want the bars lower than the touring or recreational cyclist. The cyclist who pedals all the time will be confortable with lower bars than the cyclist who coasts down hills.

The "7" shaped handlebar stem gets its shape from a historical accident. The style in the old days was to ride rather tall frames by our standards, and the older handlebar shapes had less drop than modern designs. (When the transition from the "highwheeler" to the "safety" bike occurred, the idea of being able to stand over the frame did not occurr immediately. Cyclists were in the habit of having to mount and dismount on the fly.) The stem would usually be inserted so that only an inch or less stuck out of the headset. The "7" shape allowed the lowest possible handlebar placement with as much forward reach as was wanted.

Nowadays, most cyclists set their stems all the way up (at the "minimum insertion" mark). With the smaller frame sizes used now, the "7" shaped stem is an atavism, a stylistic holdover from an obsolete technology. An extended "7" stem is two sides of a triangle. A stem that follows the diagonal, directly from just above the headset to the handlebar clamp makes more sense geometrically. Such a stem would be as strong as a similarly made "7" stem, but substantially lighter. It would also be more crash-worthy. Modern Allen-bolt stems are certainly safer than the old style that had a protruding hex head and a sharp rear corner, but the shape is still a threat to the rider's groin in a collision.

There is a trend to use "mountain-bike type" stems on road bikes, and it really makes a lot of sense. All that the "7" stem has going for it is tradition.

When you see a small frame size that features a higher bottom bracket than the other sizes of the same model, you're seeing dishonesty and deliberate deception.

This isn't nearly as common as it used to be, since the abandonment of a slavish dedication to the level top tube, but used to be widespread.

Here's how it works:

Let's say a given model comes in 25, 23, 21 and 19 inch frame sizes. A shop might have a 21 and a 23 on the floor. A small rider comes in, tries to stand over the 21, but it's too tall. She figures, well yeah, that's too tall, but the 19 ought to be two inches lower, and I'm sure I could fit a bike two inches lower, so I'll order one.

However, the 19 inch frame has a 3/4" higher bottom bracket, so the standover height is actually only 1 1/4" lower than the 21" model! Now, if the manufacturer really cared about customers who need the smaller frame size, they'd install shorter cranks on the 19". This would actually permit them to _lower_ the bottom bracket of the 19" size, since there's less of a pedal strike issue with the shorter cranks.

The problem turns out to be basically that the wheels are too large for the rider. However, there are marketing difficulties in selling a bike with smaller wheels, so that's an ignored option.

If you want to build a 19" frame with full-sized wheels _and_ a level top tube, you wind up with an itty-bitty head tube and steerer. This creates issues with the headset and handlebar stem, so they can't go below a certain minimum head-tube length.

That's where the cheating comes in...if you raise the bottom bracket, you can make a frame you can call a 19" and still have the level top tube and a reasonable head tube.

As I mentioned, this is no longer that common, since the sloping top tube eliminates the problem, but short riders looking at older used bikes should beware this scam.