Summing up, the greater proportion of neutral sites, the more rapid the rate of molecular evolution. So in accordance with the neutral theory, the rate of which genes evolve is determined by the overall rate of mutation and proportion of neutral sites. Darwin actually predicted two phenomena (rate of fixation of mutations and high level of polymorphisms), which may be accounted by the neutral theory. And the amount of divergence between genes tends to increase (with time) since their evolutionary separation. But molecular clocks themselves may vary either as a result of 'sloppiness' of the tick-rate or variation in the mutation rate; since the clock is probabilistic (ticks are irregular intervals which can be described by a Poisson distribution), but where did this variation stem from? An important source of variation comes from the influence of population size on the rate of fixation of mutation. Ohta expanded the neutral theory (with the nearly-neutral theory) by acknowledging the important role of effective population size; smaller populations are more severely influenced by fluctuations in allelic frequency, so genetic drift can vanquish selection for alleles with small selection coefficients. So in effect, the fixation of of nearly-neutral alleles of small selection effect is predicted to be the greatest in the smallest populations, if a population undergoes a decrease in population, this might coincide with an influx of fixation of nearly-neutral alleles, so population flux can increase the sloppiness of molecular clocks. Another application of molecular clocks can be made to the Hawaiian Islands, where the phylogeny of endemic birds and fruit flies is confirmed by molecular dates that follow a linear correlation between divergence and time in which DNA distance is compared against Island age. Since viruses leave behind no fossil record, we can also reassemble the history of viral outbreaks using viral lineages (viral molecular clock). In the case of endogenous retroviruses (ERVs), dates of origin can be fine tuned by comparing the pair of long terminal repeats (LTRs) that surround the genome.