Contacts between rocks recording large differences in metamorphic grade are indicative of major tectonic displacements. Low‐P upon high‐P contacts are commonly interpreted as extensional (i.e. material points on either side of the contact moved apart relative to the palaeo‐horizontal), but dating of deformation and metamorphism is essential in testing such models. In the Western Alps, the Piemonte Ophiolite consists of eclogites (T ≈550–600 °C and P≈18–20 kbar) structurally beneath greenschist facies rocks (T ≈400 °C and P≈9 kbar). Mapping shows that the latter form a kilometre‐wide shear zone (the Gressoney Shear Zone, GSZ) dominated by top‐SE movement related to crustal extension. Rb–Sr data from micas within different GSZ fabrics, which dynamically recrystallized below their blocking temperature, are interpreted as deformation ages. Ages from different samples within the same fabric are reproducible and are consistent with the relative chronology derived from mapping. They show that the GSZ had an extensional deformation history over a period of c. 9 Myr between c. 45–36 Ma. This overlaps in time with the eclogite facies metamorphism. The GSZ operated over the entire period during which the footwall evolved from eclogite to greenschist facies and was therefore responsible for eclogite exhumation. The discrete contact zone between eclogite and greenschist facies rocks is the last active part of the GSZ and truncates greenschist facies folds in the footwall. These final movements were therefore not a major component of eclogite exhumation. Pressure estimates associated with old and young fabrics within the GSZ are comparable, indicating that during extensional deformation there was no significant unroofing of the hangingwall. Since there are no known extensional structures younger than 36 Ma at higher levels in this part of the Alps, exhumation since the final juxtaposition of the two units (at 36 Ma) seems to have been dominated by erosion. Key words: deformation age, eclogite, exhumation, Rb–Sr dating, tectonic.

Sarcoidosis, a multisystem disease of unknown aetiology, is characterized by the formation of non-caseating granulomas which may involve any organ of the body. The commonest sites of predilection are the lungs, skin and lymph nodes. We now report a patient who presented with palmar erythema which on biopsy confirmed the presence of non-caseating granulomas and who responded to systemic corticosteroids. We are unaware of any previous report in the literature of sarcoidosis presenting in this way.

Abstract-If a VLSI chip is partitioned into functional units (NJ's) and redundant FU'S are added, error correcting codes maybe employed to increase the yield and/or reliability of the chip. Acceptable testing is defined to be testing the chip with the error corrector functioning, thns obtaining the maximum increase in yield afforded by the error correction. The acceptable testing theorem shows that the use of coding and error correction in conjunction with acceptable testing can significantly increase the yield of VLSI chips without seriously compromising their reliability.

A parameter estimation problem for class of linear hereditary systems i s r e c a s t i n a s t a t e space setting. A quasilinearization type algorithm is developed, based on the state space f o r m l a t i o n . The resulting algorithm requires t h e s o l u t i o n o f i n i t i a l v a l u e problems f o r hereditary systems. A numerical example i s g i v e n t o i l l u s t r a t e t h e t e c h n i q u e .

Abstract-An analog data acquisition system typically consists of an amdog multiplexer followed by an analog to digital (A/D) converter. An alternative configuration uses multiple comparators (one per input) followed by a digitai multiplexer.If the control functions for A/D conversion are incorporated into a microprocessor, several potentially interesting data acquisition and conversion strategies are available. At fiist thought, an interruptdriven conversion process which simultaneously searches all inputs in parallel for a level match appears attractive. However, analysis shows that a sequential conversion of the inputs, one at a tiWe, using the successive approximation algorithm, is usually superior. L INTRODUCTIONAnalog data acquisition is a function of many microprocessor systems. Typically the part of the system which does the analog data acquisition uses an analog multiplexer which (under microprocessor control) selects one of the analog inputs and transmits it to an analog to digital (A/D) converter. The selected analog input is digitized by the A/D converter and the digitized value of the analog input is transmitted to the microprocessor [ 1] . This method of analog data acquisition is perfectly acceptable for most applications.There are, however, some problems associated with the analog multiplexer, which normally consists of an array of analog semiconductor switches [2] . It is the purpose of this correspondence to explore the ramifications of doing the multiplexing function in the digital domain and of using the microprocessor to replace the hardwired c'ontrol logic which resides in the typical A/D converter. II. ELIMINATION OF ANALOG MULTIPLEXORA/D conversion can be accomplished in many ways [ 1] . At the black box level, an A/D converter partitions the analog input space into n contiguous disjoint subspaces.It further makes a decision as to which of these subspaces the current value of the input variable resides in. It then outputs a digital code which uniquely identifies that sub space.If the input signal could change significantly in the time required to do the A/D conversion, then it would be necessary t o include a sample-and-hold circuit to freeze the signal level applied to the A/D converter.However, if the signal can not change significantly during the conversion time (a thermistor output for instance), then the sample-and-hold is not required [31.One way to eliminate analog multiplexing is to employ a separate A/D converter for each analog input [3] . Indeed for some systems (for instance if there are relatively few widebandwidth inputs) this may be the most viable approach. However, for systems with large numbers of narrow-bandwidth inputs (such as spacecraft housekeeping data) there is a more practical method.A frequently used type of A/D converter functions by repeatedly comparing its analog input to a sequence of analog levels which are generated by a D/A converter [ 1]. At each step of this process, the output of the comparison function is Manuscript received June 8, 1979; revised No...