It's easy to see how, if G1 and G2 are separated by a long stretch of chromosome, they are more likely to become unlinked by crossing-over than if they are close together.

We can turn this relationship on its head and say that if we determine the frequency with which two linked genes become unlinked, we have a measure of their physical locations on the chromosome.

The relationship shown in this graph makes the point but is, of course, a gross over-simplification.  If you're interested, take a look at this, and read more here.

This is the basis of the powerful technique of genetic mapping, established a century ago by Thomas Hunt Morgan's group working with the fruit fly Drosophila.  In Morgan’s honour, the unit of genetic distance is called the centimorgan (cM).

In our imaginary example, let’s say that G1 is a gene that controls the rate at which chloroplast protein is broken down during senescence.  G1 red might be a variant of the gene that causes fast breakdown and G2 blue is a slow breakdown variant (variants of a particular gene are called alleles).

And, for the sake of argument, let’s imagine that G2 red and blue are fast and slow alleles respectively of a gene controlling rate of chlorophyll breakdown.  Finally G3 red and blue are responsible for, say, high and low respiration.

We make a cross between parents differing in rates of protein and chlorophyll breakdown and respiration.  We then measure these things in individuals of subsequent generations to record how the high and low variants of each character segregate during inheritance.  This picture shows leaves from individuals in a ryegrass population segregating for slow and fast chlorophyll breakdown.

By working out the proportions of all the combinations of trait variants we can determine which of the corresponding genes are in the same linkage group (that is, on the same chromosome) and what their relative distance apart is.

So in our example we will find that G1 and G2 are in the same linkage group, but G3 is in a different group.  We would therefore infer that our imaginary plant has a chromosome complement of at least n = 2.  Moreover, we can get an estimate, in cM, of the distance on linkage group (that is, chromosome) 1 between G1 and G2.

In principle, this is how linkage maps have been worked out for organisms from yeast to flies to moss to frogs to crop plants to humans.  Here is a genetic map of pea:

And here's a map of the seven chromosomes of ryegrass (Lolium perenne), showing how gene order relates to the map of the twelve chromosomes of rice: