There are these fossilized embryos from the Ediacaran, approximately 570 million years ago, that have been uncovered in the Doushantuo formation in China. I’ve mentioned them before, and as you can see below, they are genuinely spectacular.
But, you know, I work with comparable fresh embryos all the time, and I can tell you that they are incredibly fragile—it’s easy to damage them and watch them pop (that’s a 2.3MB Quicktime movie), and dead embryos die and decay with amazing speed, minutes to hours. Dead cells release enzymes that trigger a process called autolysis that digests the embryo from within, and any bacteria in the neighborhood—and there are always bacteria around—descend on the tasty corpse and can turn it into a puddle of goo in almost no time at all. It makes a fellow wonder how these fossils could have formed, and what kind of conditions protect the cells from complete destruction before they were mineralized. Another concern is what kinds of embryos are favored by whatever the process is—is there a bias in the preservation?
Now Raff et al. have done a study in experimental taphonomy, the study of the conditions and processes by which organisms are fossilized, and have come up with a couple of answers for me. Short version: the conditions for rapid preservation are fairly easy to generate, but there is a bias in which stages can be reliably preserved.