Presented in this paper is a robust controller design methodology for a class of uncertain, multivariable, regulating systems required to maintain a prespecified operating condition within hard time domain tolerances despite a vector of step disturbances. The design methodology is a frequency domain approach and is based on sequential loop design where a Gauss elimination technique facilitates the various design steps. The specific class of systems addressed are those which can be modeled as square, multivariable systems with parametric uncertainty. One restriction imposed is that the system and its inverse are stable for all plant parameter combinations. The key features of this design methodology include (i) the design of a fully populated controller matrix, (ii) the ability to design for system integrity, and (iii) the direct enforcement of hard time domain tolerances through frequency domain amplitude inequalities.

Abstract.-A study of the hearing of the penguin Sphenikcus demersus, in terms of the cochlear potentials, showed sensitivity over a range at least of 100 to 15,000 Hz, with the best sensitivity in the region of 600 to 4000 Hz. The form of the sensitivity function is consistent with the vocalizations of these animals. In general, this species of penguin shows good agreement with other birds in both the form and range of auditory sensitivity.The auditory capabilities of one species of penguin, Spheniscus demermus, commonly called the Blackfooted or Jackass Penguin, were investigated by means of the cochlear potentials. This is one of the smaller penguin species, breeding on islands off the western coast of South Africa. Three animals were used for the observations.The animals were anesthetized with ethyl carbamate, administered intraperitoneally in a 20 per cent solution at a dosage of 0.01 cc per gram of body weight. During the experiments the animal rested on a heating blanket, and a body temperature of 380C was maintained by means of a temperature probe in the cloaca that led to a regulating instrument.The middle ear cavity was exposed by a lateral approach, as indicated in Figure 1. The right side of the head is shown, with the region of the skin incision indicated. The removal of skin and muscle tissue in this region exposed the poster-

A single-lever shock avoidance conditioning technique was used to generate auditory behavioral thresholds for Lemur catta over the frequency range 100 Hz to 75 kHz. The results indicate a sensitivity of at least 60 dB below one dyne/cm2 from 1 kHz to 32 kHz and, throughout the entire frequency range, appear to be compatible with other available prosimian data. A contrast between anthropoids and prosimians indicates a loss in high frequency sensitivity for the former group, but a greater sensitivity for lower frequency stimuli. Analysis of lemur vocalizations indicates a preponderance of relatively low frequency components. Descriptions of the conditioning technique, threshold criteria, and procedures of sound measurement are included.As part of a long-term investigation of the hearing of prosimians, we have assessed the auditory capabilities of several species of lemurs by means of conditioning techniques. We chose this representative of the suborder prosimii since no quantitative behavioral data on hearing were available for this animal (Vernon, '67) and the Oregon Regional Primate Research Center maintains one of the largest captive populations of lemurs in the world.In a previous paper (Mitchell, Gillette, Vernon and Herman, '70), we reported on the auditory sensitivity of Lemur catta, L. fulvus, and L. macaco over the frequency range of 100 hertz to 40 kilohertz, using a two-lever positive reinforcement technique patterned after that used by Gourevitch and Hack ('66). The derived data were obtained only after long and exasperating training and testing sessions; we feel this technique is too difficult for the lemur. Further, it did not permit an appropriate degree of stimulus control, resulting in heightened variability and lessened confidence in the data. Accordingly, we adopted a different behavioral technique, similar to the single-lever shock avoidance paradigm reported by Clack and Herman ('63 ), and replicated these threshold determinations, this time considering only the species L. catta. These data (Mitchell, Vernon and Herman, '71) indi- AM. J. PHYS. ANTHROP., 38: 365-370cate an increased sensitivity, relative to the first study, of up to 20 decibels, especially for the mid-range frequencies. Variability of behavior was also considerably lessened. Since differences between species were negligible in the first study, and because we feel the 20 dB difference between the first and second study is attributable to the response contingencies of the two methods, we would accept the data of the second study as being more accurate.The present report (study 3 ) considers the high frequency sensitivity of L. catta from 8 kHz to 75 kHz; to complete the audiogram, we are also presenting the results of study 2, which extended down to a frequency of 100 Hz. METHODThe calibration procedures used in the present study differed from those previously employed; however, the training procedures, threshold testing techniques, and criteria were the same. Although a more complete description of these procedures appear...

Absolute thresholds were determined for pure tones of frequencies 100 Hz-40 kHz by the method of constant stimuli using a two-bar operant technique. Lemurs of three subspecies were trained to press a white bar to turn on a tone lasting 5 sec. During the tone, a single press on a second (black) bar resulted in food reward. Failure to respond simply terminated the trial. Response in the absence of tone was scored as a false positive response and resulted in a «-sec shock of from 0.6 to 1.3 mA. Thresholds, based on a response probability of 0.5, indicate a lesser sensitivity in the low and middle frequencies than found in the anthropoid primates, but an apparent extension of sensitivity into the higher frequencies beyond the upper-frequency cutoff points found for the anthropoids. I. PROCEDURE A. SubjectsSix lemurs, males and females of each of three species (lemur catta, lemur macaco fulvus, and lemur macaco macaco), served as subjects. Their weights, breeding behavior, and dentition indicate that three were adult and three adolescent (see Table I). All were healthy; one (No. 2023) was extremely fat and had also been previously treated with antibiotics for a jaw infection. The animals were deprived of food and received ad lib water during training and testing. The extent of deprivation may be estimated from the weight data included in Table I. B. ApparatusThe animals were trained and tested in a small cage inside a double-walled IAC chamber. Standard audiometric and programming equipment was used to generate and control the sound stimulus, food, and, later, shock. A Western Electric 555 speaker produced