Increasingly, hominins have survived swings of climate by their wits and by chance. Neither underpin the instinct to migrate when times are hard, but where one ends up depended, until the Holocene, more on chance than design. Early migrations must have been more by diffusion than purposeful, especially in the vastness of the African continent. Yet groups of hominins found their way into Eurasia several times and thrived there. Far more of them would have met the coast far from a continental exit route, such as the Levant or the Straits of Bab el Mandab. However, in stressful glacial episodes reaching the coast was a key to survival as its food resources are almost limitless (see Human migration and sea food May 2000 issue of EPN). Our own species found refuge by the sea not long after we originated (Marean, C.W. 2010. When the sea saved humanity. Scientific American, v. 303 (August 2010), p. 40-47). Around 195 ka climate began to cool and dry to reach a glacial maximum at roughly 123 ka. Curtis Marean (Arizona State University, USA) was one of the first scientists to look for signs of coastal refuges in Africa during the early 1990s, particularly at its southern tip. With co-workers he found several caves on the coast of South Africa that yielded the evidence on which he has based a review of littoral survival opportunities and the skills that we developed. This particular coastal stretch has a huge diversity of plant life, most unique to it, and many of which store carbohydrate in tubers, bulbs and corms. They are adapted to dry conditions and need only the simplest technology – digging sticks and fires for cooking – to exploit starchy, easily digested energy resources, along with the more obvious animal protein sources present on all shorelines. Marean’s review puts in plain language all the discoveries made by his group over the last 20 years, including evidence of the use of fire treatment to improve flaked stone tools and the development of art based on iron-oxide pigments, plus his own take on their anthropological significance.