Why do gaps exist? (or seem to exist)
Ideally, of course, we would like to know each lineage right down to the species level, and have detailed species-to-species transitions linking every species in the lineage. But in practice, we get an uneven mix of the two, with only a few species-to-species transitions, and occasionally long time breaks in the lineage. Many laypeople even have the (incorrect) impression that the situation is even worse, and that there are no known transitions at all. Why are there still gaps? And why do many people think that there are even more gaps than there really are?
Stratigraphic gaps
The first and most major reason for gaps is "stratigraphic discontinuities", meaning that fossil-bearing strata are not at all continuous. There are often large time breaks from one stratum to the next, and there are even some times for which no fossil strata have been found. For instance, the Aalenian (mid-Jurassic) has shown no known tetrapod fossils anywhere in the world, and other stratigraphic stages in the Carboniferous, Jurassic, and Cretaceous have produced only a few mangled tetrapods. Most other strata have produced at least one fossil from between 50% and 100% of the vertebrate families that we know had already arisen by then (Benton, 1989) -- so the vertebrate record at the family level is only about 75% complete, and much less complete at the genus or species level. (One study estimated that we may have fossils from as little as 3% of the species that existed in the Eocene!) This, obviously, is the major reason for a break in a general lineage. To further complicate the picture, certain types of animals tend not to get fossilized -- terrestrial animals, small animals, fragile animals, and forest-dwellers are worst. And finally, fossils from very early times just don't survive the passage of eons very well, what with all the folding, crushing, and melting that goes on. Due to these facts of life and death, there will always be some major breaks in the fossil record.
Species-to-species transitions are even harder to document. To demonstrate anything about how a species arose, whether it arose gradually or suddenly, you need exceptionally complete strata, with many dead animals buried under constant, rapid sedimentation. This is rare for terrestrial animals. Even the famous Clark's Fork (Wyoming) site, known for its fine Eocene mammal transitions, only has about one fossil per lineage about every 27,000 years. Luckily, this is enough to record most episodes of evolutionary change (provided that they occurred at Clark's Fork Basin and not somewhere else), though it misses the most rapid evolutionary bursts. In general, in order to document transitions between species, you specimens separated by only tens of thousands of years (e.g. every 20,000-80,000 years). If you have only one specimen for hundreds of thousands of years (e.g. every 500,000 years), you can usually determine the order of species, but not the transitions between species. If you have a specimen every million years, you can get the order of genera, but not which species were involved. And so on. These are rough estimates (from Gingerich, 1976, 1980) but should give an idea of the completeness required.
Note that fossils separated by more than about a hundred thousand years cannot show anything about how a species arose. Think about it: there could have been a smooth transition, or the species could have appeared suddenly, but either way, if there aren't enough fossils, we can't tell which way it happened.
