The first wholly acceptable association of human remains, ancient artifacts, and extinct animals was found in 1953. On the Scharbauer Ranch, near Midland, Texas, parts of a skeleton of a 30-year-old woman were recovered from a deposit of gray sand that included chipped stone tools, remains of extinct horse and bison, and the burned bone of an extinct antelope that Midland man may accidentally have dug up from a lower layer. The gray sand provided a number of materials that have produced dates by radiocarbon and other methods. A higher and younger layer of red sand contained artifacts known elsewhere to have a maximum age of about 10,000 years. This sand was laid down in a different, more humid climate. Snail shells from a layer of white sand, below the gray, produced a radiocarbon date of less than 15,000 years. Logically, Midland man should date between these extremes. But pieces of caliche rock from the gray layer, burned presumably in cooking fires of Midland man, provided a radiocarbon date of about 20,000 years. This was not acceptable, since a lower deposit seemed to be younger. There are at least two potential sources of natural dilution of radiocarbon that could help the burned caliche to test older than it actually was.
Tenuous support for the 20,000-year date came from a new source. Three samples from the gray sand were submitted for testing by a new approach to dating archaeological materials—a refinement of the “uranium-clock” technique by which very old rocks are dated. This, the so-called Rosholt method, is a measurement of the amounts of uranium and its products in radioactive disintegration that accumulate in bone or other material. A ratio is determined between two isotopes of uranium (U²³⁸ and U²³⁵) and such radioactive related products as protactinium (Pa²³¹), thorium (Th²³⁰), and radium (Ra²²⁶). The method is yet unproved and may have more sources of contamination or error than radiocarbon. The results of Rosholt measurement on Midland samples averaged 17,000, 18,000, and 20,000 years. The latter date was from a small section of the Midland skull itself. It must be reemphasized that this process is not established well enough, and the results here are not acceptable to the archaeologists who conducted the excavations.[42]