No abstract

The last decade has seen the modest application of geophysical methods to archaeological problems. Various techniques have been tried including magnetic, seismic and electrical resistivity measurement. Of these, the first has proven most useful in temperate Europe because the measured phenomenon does not depend on temporal climatic variation. The sites in this part of the world are usually not deeply buried, and the features which are sought are frequently o f simple geometric shape. These include walls, ditches, pits, kilns, graves, etc.Since 1961, large scale magnetic surveys have been undertaken in the Rhineland, using a differential proton magnetometer which was constructed at the Rheinisches Landesmuseum especially for the types of sites mentioned (Scollar 1965). One magnetic detector sonde is placed at a fixed point. The other is moved, meter by meter, over the entire surface to be explored. Values of differences in the Earth's magnetic field relative to the fixed point are recorded with an accuracy of either one part in fifty thousand or one in two hundred thousand. Where the natural soil sequence has been disturbed by man in the past, magnetic particles from the upper layer are incorporated in the subsequently buried archaeological features (Le Borgne 1965). Slight field anomalies can be measured in many cases.It is sound practice, in the relatively unencumbered flat terrain in the Rhineland, to divide up the area to be surveyed with a square grid whose corners are fixed by wooden pegs. Using the instrument mentioned, grid crossings are spaced at 20 meter intervals. Two parallel non-magnetic twenty-five meter tapes are placed at the sides of each square. Two additional cross tapes, running in opposite directions, provide a reasonably exact reference for every measured point. Usually one measurement is made per meter. This is sufficiently fine to delineate most archaeological features of interest without raising the total number of readings to an excessive degree. Thus, a hectare of surface requires at least 10.000 readings. This does not include any overlap at the sides of adjacent squares. For checking purposes described below, each square is measured completely so that all interior edges in the complete grid are effectively measured twice and all interior corners are measured four times. Since a full square takes only about twenty-five minutes to traverse using automatic recording equipment, and only about ten percent more time is needed, this repetition is not disadvantageous. 441 readings are obtained for each 20 meter square or 11,025 per hectare. A typical site in the Rhineland requires a survey of from two to ten hectares so that fifty thousand readings are not unusual. A survey of this type represents three weeks work for three men.It is immediately apparent that a large mass of data of this kind is well suited to computer evaluation. In the ideal case, this takes the form of the production of a pictorial image of the variations in magnetic field strength which can be manipulated so as to increase...

No abstract

No abstract

--ZusammenfassungAn Interval Algorithm for Solving Systems of Linear Equations to Prespecified Accuracy. We describe an interval arithmetic algorithm for solving a special class of simultaneous linear equations. This class includes but is not limited to systems A x = b where A and b have integer entries. The algorithm uses fixed point arithmetic, and has two properties which distinguish it from earlier algorithms: given the absolute accuracy a desired, the algorithm uses only as much precision as needed to achieve it, and the algorithm can adjust its own parameters to minimize computation time. AMS Subject Classifications: 65G 10 (primary), 65 F05 (secondary).Key words. Systems of linear equations, interval arithmetic, fixed point arithmetic. Genauig-keit. Wir beschreiben einen Intervallalgorithmus, der eine gewisse Klasse yon linearen Gleichungssystemen 16st. Diese Klasse enthNt u. a. Systeme A x = b, bei denen A und b ganzzahlige Komponenten haben. Dieser Algorithmus verwendet Festpunktarithmetik und unterscheidet sich von friiheren Algorithmen wie folgt. Erstens: Bei Vorgabe der gewiinschten absoluten Genauigkeit a des Ergebnisses ben6tigt der Algorithmus nur so viel Zwischengenauigkeit wie notwendig, um die Fehlerschranke a zu erreichen. Zweitens kann der Algorithmus selbststeuernd seine eigenen Parameter dynamisch/indern, um die Rechenzeit zu minimieren. Ein Intervallalgorithmus fiir die L~sung yon linearen Gleichungssystemen mit vorausgewiihlter