A reinvestigation of the middle Silurian Rose Hill Formation yields a revised pole position of 19.0 ¿N, 129.1 ¿E (α95=5.8¿ for N=35 samples from nine sites and three lithologies). This is based upon three separate fold tests, all of which indicate predeformational stability of the magnetization at the 99% confidence level, as well as upon normal and reversed polarity directions. Laboratory methods used to derive these results included thermal, chemical, and alternating field stepwise demagnetizations of three lithologies: gray sandstones, hematitic red sandstones, and dolomitic sandstones. Consistency among the techniques was excellent, and through vector subtraction we obtained characteristic and secondary components of magnetization. Reversed polarity directions of the hematite sandstones did not produce a positive fold test, and their resemblance to late Paleozoic directions for North America suggests that they have been remagnetized during that time. The remagnetization was probably a chemical event, since a thermal event would have also remagnetized the gray sandstones and the dolomitic sandstones. In addition to paleomagnetic experiments, rock magnetic experiments were performed to clarify the nature of the dolomites (previously called siderites). Analysis of natural and isothermal remanent magnetizations from96 ¿C to 680 ¿C shows a dominance of fine-grained magnetite and hematite. Thin sections show that lesser amounts of coarse-grained magnetite and hematite are also present. Consistency between thermal, chemical, and alternating field demagnetizations indicates that both hematite and magnetite carry the characteristic direction. The mean inclination for the Rose Hill Formation, -44¿, places North America in moderate southerly latitudes (about 26¿) during the Middle Silurian. The declination of 330¿ indicates a rotation of 30¿ in a clockwise sense from the present-day orientation, since we cannot determine if relative rotations of blocks in this part of the Appalachians occurred, or whether the declination is truly representative of the entire North American craton. However, evidence for horizontal rotations on the scale of 1 m to 100 km is lacking, so we prefer a rotation of the entire North American craton by this amount.