The human embryo at around day 10–11.

      The fate of the embryo wholly depends on a successful implantation in the lining of the uterus. If this does not happen – which is the case around 50 per cent of the time – the embryo will be expelled in the menstrual cycle. Failure to implant is one of the common reasons why an IVF cycle does not work. No wonder, then, the division of labour in the blastocyst makes it seem that its priority is to those cells surrounding the epiblast, which won’t be a part of the fetus at all. For without implantation, it’s game over.

      Implantation is a delicate and complex process involving a dialogue of hormones and proteins between the embryo and the cells of the uterine lining. In some ways it is more delicate and complex than fertilization itself. The placenta, for example, is made not just from the trophoblast layer of the blastocyst but also from tissues from the mother, called the decidua. The two types of cell, with different genetic makeup, have to work together to create a single, vital organ. Emotive and anthropomorphic metaphors suggest themselves, presenting implantation as an intimate collaboration between the tissues of mother and her “child”. But one might equally choose to speak of the blastocyst “invading” the uterine tissue: one “organism” colonizing another for its survival.⁹ Both are stories; neither is a neutral description of events (which story ever is?).

      * * *

      The best is about to come. Calling the part of the embryo fated to become the baby an “inner cell mass” is no euphemism: it really does seem to be a shapeless conglomerate. If we want to insist that baby-making is a miracle, what seems truly miraculous is not just that the inner cell mass makes a body but that, most often, it makes exactly the same type of body, with five fingers on each hand, with all facial features in the right place and fully functional, and with its battery of correctly positioned organs. It’s no surprise that development of the embryo occasionally goes awry; it is astonishing that it does so rather rarely.

      When embryos start off as single cells, they have no plan to consult. Cells are programmed to grow and divide, but it isn’t meaningful to think of a human being as somehow fully inherent in a fertilized egg, any more than one can regard the complex convolutions of a towering termite mound as being programmed into each termite. The growth of an organism is a successive elaboration of interactions within and between cells: a kind of collaborative computation whose logic is obscure and convoluted, and the outcome of which is incompletely specified and subject to chance disturbances and digressions.

      In this way, the job evolution has devised for those formative cells is an architectural one: a challenge of coordination in time and space. They have to move into position, to acquire the right fate at the right time, and to know when it is time to stop growing or to die.

      Developmental biologists talk of this as “self-organization”. It could make the process sound quasi-magical, calling as it does upon the image of the cell as an autonomous being with aims and purposes. But many of the rules are now broadly understood.

      Two key factors are at work. First, as the cells divide and multiply, they take on increasingly specialized roles, a process called differentiation. Thus, totipotent cells in a two or four-cell embryo become trophoblasts or the pluripotent stem cells of the epiblast. The latter go through further stages of differentiation that ultimately produce the specialized cell types found in muscle, skin, blood and so forth. We will see shortly how that happens.

      Second, particular spatial arrangements may arise from cells actively moving through or across the growing organism or organ, or becoming sorted into clumps of different cell types by preferential stickiness, often between cells that are alike.

      That cells have adhesive qualities joining them into aggregates was suggested in the 1890s by Wilhelm Roux. He was also able to disrupt frog embryos by vigorous shaking, which separated them into single cells. He found that those cells would join back together, which he attributed to some kind of attractive force.

      Such “disaggregation” experiments were taken further in the early 1900s by marine biologist Henry V. Wilson, who found that sponges kept for a long time in an aquarium became “loose” and could be teased apart into individual cells. He achieved the same thing in fresh sponges by the simple measure of squeezing them through a piece of silk, which acted as a sieve that separated the cells. Again, those cells would reassemble if brought into contact to regenerate a living sponge. It was like a recapitulation of the evolution of primitive multi-celled organisms from colonies of single-celled ones (see the First Interlude, here). When Wilson did the experiment with different species of sponge, he found that cells from the same species would stick together selectively. Ernest Everett Just discerned in the 1930s that the reason for this selectivity had something to do with the cell membranes. The truth is that cells adhere via protein molecules protruding at their membrane surface (especially those belonging to the class called cadherins), which will bind to one another discerningly.

      This notion of “tissue affinity” was developed around the same time by the German-American embryologist Johannes Holtfreter. In 1955, he and Philip Townes studied how the cells of amphibian tissues that had been disaggregated by exposing them to alkalis could reassemble from solution. Holtfreter largely outlined the concept of cell sorting that allows tissues of several cell types to adopt particular structures and arrangements.

      The process of body formation (morphogenesis) is orchestrated by genes, and no wonder then that genes have been attributed such determinative power. Some researchers have made more apt comparisons to a musical score: genes tightly constrain but do not fully prescribe the performance. This is still a limited metaphor, because you can look at the score and figure out (if you’re a musician) pretty much how things will go. Not so with genes. Sometimes it is better simply to tell the story as it is, as simply as you can, rather than trying to pretend it is some other story.

      Morphogenesis literally means shape-formation, but equally it is a question of cell specialization: the embryonic stem cells gradually lose their versatility as they divide, becoming geared instead to do the task of specific tissue types. Heart muscle cells must execute synchronized beating, pancreatic cells