|MadSci Network: Genetics|
First let's go through what linkage means. Ignoring for a moment chromosomes and DNA, the concept of a gene is simply a unit of inheritance. Most higher organisms are diploid, meaning that we all carry two copies of each gene (each with its own trait), only one of which is passed on to our children. Under normal conditions, if you take a complete collection of genes, called a genome, and randomly divide the copies of the genes into new genomes, then for any given trait, the probability of also getting another trait is around 50%, i.e. for any trait the probability of being passed on to the new genome is 50%. In an absolute sense, if two genes are "linked", it means that if a trait for one gene is passed on, then a specific trait for the other gene will be passed on 100% of the time, i.e. the genes are physically linked together: inseparable. In reality, two genes are "linked" if the probability of inheriting both traits is significantly greater than 50% - there actually no 100% linkages, unless the two traits are from the same gene. To address the second part of your question, here's a little history.
While working as an undergraduate student in Thomas Morgan's lab, Alfred Sturtevant realized that the linkages his boss had been cataloguing could be rearranged to form a "linkage map" giving the positions of the genes along a chromosome. To make the map, Sturtevant defined the distance between two genes that would recombine once in every hundred matings (i.e. a 1% recombination frequency) as a centiMorgan (cM), now more commonly referred to as a genetic map unit (m.u.).
Put more simply, if we assume that recombinations occur randomly during meiosis, then genes that are closer together on the chromosome should be less likely to be separated than genes that are far apart. Some of this goes back to Gregor Mendel, who noticed that some of the traits he was studying were mutually independent (i.e. having one trait didn't affect the appearance of the other trait), while other traits were linked (i.e. having one trait determined the appearance of the other trait). Actually, Mendel fudged his data (gasp!) to suggest that linkage was absolute and without recombination - something about proving the immutability of nature. Instead, later geneticists began cataloguing the extent of linkage between traits, and grouped the genes into linkage pools even before anyone had figured out what chromosomes were for. With the identification of chromosomes as the physical manifestations of the genome, it became important to apply the linkage pools to the chromosomes.
So here's how it works. Say we have three traits that are linked, and after several hundred matings (this doesn't work well for large mammals) we determine that: in 2% of the matings, A separated from B; in 1% of the matings, A separated from C; and in 3% of the matings, B separated from C. From these we can convert to map units and say that A - B is 2cM, A - C is 1cM, and B - C is 3cM, or:
along the chromosome.
Try the links in the MadSci Library for more information on Genetics.