The behavior of a metal vapor plasma arc in a vacuum arc remelting (VAR) furnace is believed to contribute to the formation of defects in reactive metal and super-alloy ingots. Industry standard instrumentation, which includes electric current and voltage measurements, can asses the stability of an arc but cannot predict the location of an arc. It is known that Maxwell's equations predict a magnetic flux density at a distance from an arc. It is shown that a single arc's location can be uniquely determined in a cross section by using an externally mounted 2-axis Hall Effect magnetic flux density sensor provided that the system's electric current is also measured and the geometry of the VAR furnace is known. The solution is based on the Biot-Savart Law with finite element modeling assisting the analysis. The methodology is validated using controlled, static experiments. The measurement system is deployed on a small scale, experimental VAR furnace to investigate arc behavior. Results from VAR operation show a time averaged arc distribution that does not significantly change over the course of a melt. By comparing the results from multiple sensors, observed arc motions are categorized as being either retrograde or sympathetic. The former is characterized by large periodic motions, and the latter either small random motions or motions associated with an event such as a liquid metal drip short. Significant alternating currents are found to exist in the DC VAR furnace. A magnetostatic single arc model is not sufficient to describe the current distribution in the VAR furnace at an instant but it may be an effective means to detect quasi static non-axisymmetry or slow time varying current profile changes during VAR operation.