I'm trying to understand how these things work...it only may be relevant in that if there is gradual change in people according to their environment, eventually certain traits may be completely eliminated?
Well, both yes and no. Random mutation could always have a chance of reintroducing a trait (hence why certain deadly conditions that are carried by dominant genes and kill before reproductive age can persist), though it will be unlikely to ever become as prevalent as it once was through natural selection if all the people that expressed the trait suddenly died.
Whenever a population is brought near extinction, genetic variety suffers, and certain genes are lost to the population. However, in natural selection, there is no intent, so the genes lost could be benign or harmful or neutral, just by chance of what parts of the population die. Over time, populations can and do recover from these genetic bottlenecks if they survive long enough, and genetic variation will increase back to what it used to be. This, however, takes many, many generations. Cheetahs went through an extreme genetic bottleneck about 10,000 years ago, such that all modern cheetahs are descended from fewer than a dozen survivors. To this day, all cheetahs are compatible organ donors with each other, because they lack genetic diversity to such an extent.
As long as there is a particular gene still lingering about so to speak (is it recessive genes I'm thinking about, rather than dominant?), could that rematerialise even thousands of years after it was more common?
Technically, yes, and you are thinking of recessive genes. It would be unusual for a freely breeding population to need that much time, even with the long human generations, but it could happen, hypothetically.
Supposing for example, half the people in a particular population had blond hair and half had black hair, but for some reason for decades or centuries or millennia, no-one was born with black hair...would it be possible for someone to suddenly be born with black hair, and everyone would be completely taken by surprise? Could that gene remain dormant, so to speak, for that long?
Well, black hair is the dominant gene of the two, but if you applied this situation as the people with blond hair dying out, it would still be possible for a blond haired child to be born, even after generations of it not happening. It'd just be unlikely to take that long, unless the majority of black haired carriers of the blond hair genes also died. Furthermore, thanks to random mutation, it would be possible for a blond haired person to be born even if all people with the genes for that hair color died out some time before, it certainly wouldn't be guaranteed, though.
Sorry, not explaining my question well, but am trying to understand the relationship between what is inherently present in genetic material and what might actually gradually change (mutate?) or be lost, but without other human contact from another population, so that certain traits would become more or less 'fixed'..the new 'normal' if you like.
The only genes that almost never change are the hox genes, which control how developing embryos grow and develop their bodies in the womb or egg. This is because most changes to these genes would result in nonviable embryos that would die. Thanks to that, they are very much retained and similar even between species that aren't closely related, such as humans and sharks.
However, no gene is immutable, for even a mutation in a hox gene can be neutral or benign, it's just relatively uncommon. How likely a gene is to experience a mutation has much more to do with it's position on the DNA strand than it's role, with genes located more towards the center being more protected and mutating less than those more towards the ends. No trait is fixed. Nothing is inherent either, not even the nucleotide bases of DNA itself, as there are a number of conditions associated with one of the usual 4 bases being replaced by a nucleotide not usually associated with DNA, such as uracil, which is a typical component of RNA.
Furthermore, so many codons (units of base pairs that code for an amino acid) are redundant and do the same thing that I could artificially cause mutations in more than a tenth of the genes of a developing human embryo and still have said embryo grow up to appear as a normal human, and even be capable of reproduction. Most mutations do pretty much nothing, because most occur on parts of DNA that aren't even genes or are redundant changes to genes.