When we discussed traits, we described a bicycle's blue color as an example of a trait. We said that the trait (color) was produced by a single instruction in the bicycle's assembly manual. But we know that all bicycles aren't blue, so there must be other instructions that can substitute for "Paint the bicycle blue" in their assembly manuals. Other must say "Paint the bicycle red" or green or some other color. When describing man-made objects, we call these possible alternate instructions "options" or "choices". In living things, each of the possible Instructions is called an allele. For example, human beings have three alleles for alcohol dehydrogenase I. The three workers produced by these alleles differ in how quickly they can detoxify ethyl alcohol, leading to three different traits for how much and how quickly a person can drink alcohol without getting drunk. The variety of people's eye colors is due to people having different alleles as well.

When eukaryotes have more than one allele for an Instruction, their Instruction Books have two possibilities for what sort of alleles they contain. Either both Books have the same allele, or the two Books have different alleles. For example, people have two alleles for eye color. One allele produces blue eyes, the other brown. People with one of each have green or hazel eyes. The mixed colors are produced for the same reason that mixing paints produces new colors -- people with one Instruction for the blue-dye-maker and one for the brown-dye-maker make some of each, so their eyes end up with a little of both colors, blended together. The alleles for alcohol dehydrogenase work this way too: if a person has two different ones, the rate at which they metabolize alcohol is the average of the rates of the two different workers.

Not all differing alleles produce "blending" traits -- sometimes, the trait the organism shows is due to one or the other of the alleles. In the case of sweet peas, there are two alleles for flower color: pink (mentioned before) and white. The white allele is caused by an Instruction that makes a worker that can't make the pink dye. Each sweet pea plant contains one of three possible sets of Instructions for flower color-makers: two for pink, one for pink and one for white, or two for white. In sweet peas, however, there are only two traits -- pink flowers or white flowers -- because the presence of at least one Instruction for a functional dye-maker produces pink flowers. So, when looking at sweet peas, it's not possible to tell if they produce pink flowers because of one or two copies of the pink allele. Traits like this, where one allele takes precedence over another rather than blending, are called Mendelian (after Gregor Mendel, who discovered them). The allele that takes precedence is called dominant and the one that disappears (or recedes into the background) is called recessive.

Whether organisms with two different alleles for the same Instruction in their Books have Mendelian (like flower color) or "blending" (like eye color) traits depends on the workers and how they interact. There aren't any simple rules that say which must happen.

The interaction of two different alleles may produce traits that aren't like any Instructions in the living thing's Instruction Books. People have green eyes even though there isn't an Instruction that makes a green-dye-making worker. Pea plants can be pink even though one of their books has an Instruction for a worker that doesn't work. This difference is important to people who want to know what kind of offspring a living thing will produce, like farmers breeding animals or doctors trying to see if a person might pass on a genetic disease. For example, consider the case of a pea plant with one allele for pink flowers and one for white. While this doesn't matter to much to the plant itself, it matters to its offspring: if the plant has two pink alleles, when it reproduces, all of its offspring will get a pink allele, since the pink allele dominates the white one. However, if it has one pink and one white allele, half of its offspring will get the white allele, and those plants' color will be determined by the allele they get from their other parent, since the white allele is recessive. People who have to distinguish between the Instructions a living thing's Instruction Book contains and the traits the living thing displays use two different words to keep track of them. The collection of Instructions in the organism's Instruction Book is its genotype and the traits the organism shows are its phenotype.

The difference between genotypes and phenotypes is the key to understanding the transmission of human genetic diseases. Many genetic diseases are caused by an allele that produces a broken worker that can't do its job. Hemophilia (an inability to clot blood and form scabs), phenylketonuria (an inability to properly metabolize some amino acids), and Tay-Sachs disease (an inability to break down a molecule in nerves) are like this. The alleles for these genetic diseases are recessive. If a person has one Instruction Book with a broken-worker allele, and their other Instruction Book contains an allele for a functional worker, they suffer no effects from the broken one. These people have normal phenotypes, but their genotype contains one broken-worker allele. Since these people carry an allele for the disease, but are otherwise normal, they're called carriers of the disease.

In general, being a carrier for a genetic disease isn't all that bad. Not only do carriers not suffer from any effects, their children aren't likely to either. Why? Because most genetic diseases are rare, so most people they'd have children with have two functional alleles. When a carrier has children with a normal person, each of their children has a fifty-fifty chance of being normal (two functional alleles) and a fifty-fifty chance of them being carriers (one functional allele and one broken-worker allele). However, when two carriers have children, the situation is different. Each carrier has a fifty-fifty chance to pass either of their two alleles on, one of which produces broken workers. As a result, each child has a 25 percent chance to be normal (with two functional alleles), a 25 percent chance to suffer from the genetic disease (because of two broken-worker alleles), and a 50 percent chance to be a carrier (with one of each allele). Sadly, before the invention of genetic testing the only way to be sure if a person was a carrier was to produce children with the genetic disease. (Of course, if a person's near relatives had the genetic disease, they may have suspected that they were a carrier.)

Where a trait is produced by multiple Instructions, each Instruction can have multiple alleles, leading to many possible traits or many ways to get the same trait. For instance, we mentioned above that the ability to drink alcohol without getting drunk is determined by an allele for alcohol dehydrogenase I. This isn't quite right, since it turns out that the other worker in the alcohol-detoxifying pathway, aldehyde dehydrogenase, has two alleles. Since these also differ in how well the workers they make carry out their jobs, humans possess many levels of alcohol tolerance, ranging from almost none (slow varieties of both workers in both Instruction Books) to very high (fast varieties of both workers in both Instruction Books), and many combinations in between. (This is still a bit simplified, since other workers affect alcohol metabolism slightly. Other non-genetic factors, such as a person's weight, also affect alcohol metabolism.)

While we've been discussing traits and Instructions in living things as if they were traits and instructions to make manufactured goods, this isn't always the case. For example, in putting together a bicycle, we can see that certain instructions are clearly related to traits. For instance, the bicycle's color is set by the Instruction that says what color to paint it. The color Instruction doesn't change any other trait the bicycle has. In living things, however, workers produced by Instructions are often very similar, and can sometimes do the job of another worker. When we discussed specificity we mentioned that alcohol dehydrogenase III is capable of working on ethyl alcohol a little bit. Even though this isn't its normal job, it can do it. So, even if a person's two Instruction Books both contain Instructions that produce broken alcohol dehydrogenase I, he might be able to tolerate a little alcohol because he has another worker that can sort of do the job.

This sort of "compensation" is pretty rare in man-made things. However, it does show up once in while. For instance, when a factory makes a bicycle, it might be out of the 1-inch-long screws that hold the brake pads onto the brakes. The workers might be able to get away with using 1.5-inch-long screws. Although the brakes would look funny, they might work.

The differences between genotypes and phenotypes -- whether caused by the effects of Instructions with multiple alleles or by "compensating" workers and pathways -- complicate the task of trying to figure out what Instructions go with what traits. This is especially hard if one is starting with traits, and trying to figure out what Instructions might go with them. Despite the tremendous difficulty of this problem, this is what mass breeders must do when they try to improve organisms via artificial selection.

How can people improve organisms by breeding them? What part gets changed in this improvement? The answer, of course, is the organism's Instruction Book. As organisms grow and reproduce, they copy their Instruction Books. Let's take a look at how they do that. Along the way, we'll see how and why the Book changes as it's copied.