Joint often play a critical role in localizing economic deposits of petroleum and groundwater. A knowledge of reservoir fracturing can optimize resource development and increase fluid recovery. Such reservoir units are frequently found in regions with nearly flat-lying, unfaulted strata. In these areas the ability to predict vertically extensive joint sets using surface mapping can provide the competitive edge in exploration and production. It is commonly assumed that surface joints are related to joints at depth, yet until now there has been no widely accepted mechanism explaining when and how joints propagate vertically. Cyclic loading resulting from earth tides or recurring seismic shocks may provide that mechanism. The objective of this work is to determine the effect of a pre-existing joint on initiation and propagation of cracks across lithologic contacts in cyclic loading. Pikes Peak granite and Dakota sandstone were tested in uniaxial tension and compression, triaxial compression, and triaxial cyclic compression-tension. Both monolithologic and sanstone-granite composite samples were tested. Composite samples were meant to simulate the basement-sediment contact in underformed rock. Two basic configurations were examined: that representing lateral (horizontal) loading, and that representing vertical loading. The granite was saw cut to approximate pre-existing joints. Uniaxial tests provided information on elastic properties of the two rock types and the relative strength of various configurations. Triaxial tests were used to simulate lateral and vertical loading conditions under 308m (1000 ft) of overburden. Triaxial cyclic fatigue tests were used to examine the geometry of joints induced under lateral and vertical cyclic loading, the amount of weakening possible, the relative importance of compression vs tension in cyclic loading, and the relation of induced fractures to pre-existing joints. The results of these tests were compared to observations made at outcrops of basement-sandstone contacts. Given the opportunity, rock will fall in tension before it fails in compression. Rock fails by alternting lateral and axial extension in cyclic triaxial loading. Extension failure occured at as little 24% of the triaxial compressive strength during about 80 cycles. Most outcrop joints do not cross contacts, indicating that the discontinuity is a barrier to the passage of stress concentrations. The most probable cause of near-vertical joints that do cross contacts without displacement is lateral extension. Pre-existing cracks concentrate tensile stress at the contact and initiate growth of a joint from that point.