officials and others in the U. S. S. R., in Moscow. Additional evidence of PLATO's impact can be drawn from the fact that more than 500 authors (ranging in age and experience from school children to full professors) have produced more than 2000 hours of lesson material for PLATO, in more than 80 different disciplines. Although access to PLATO consoles and lessons is still quite limited, more than 20,000 student-contact-hours were logged under PLATO during the fall 1973 term. Furthermore, preliminary results from a survey of 42 instructors and about 500 students who used the system revealed that three-fourths of the instructors and two-thirds of the students felt rather positively toward PLATO CAI after one semester's use, in spite of limited access and occasional interruptions in service. There have of course been problems and dealys, as well as progress, which have resulted in a substantial increase in the cost for the elementary and community college activities and a year's postponement in the start of the field-tests. Some of this delay is attributable to problems with the hardware and software itself. Although the system reached a moderate level of performance relatively quickly, there have been interruptions in service, especially during those periods when major hardware had to be added to the system.* These interruptions were especially troublesome at the (now more than 70) remote sites, since the terminals themselves have frequently accounted for the majority of the system down-time, and many remote sites have only a single terminal. It must also be reported that the original configuration proved inadequate to serve the relatively large number of lesson authors committed to developing courseware. The initial configuration assumed that the majority of PLATO use would accrue to students enrolled in scheduled classes, who could therefore share PLATO lessons and memory space. Authors of course usually require distinct PLATO lessons and memory space, and a shortage of extended-core storage quickly developed. lt therefore proved necessary to double the amount of extended-core storage for lesson swapping. There have also been substantial delays in obtaining consoles and telecommunication equipment and service. Difficulties in obtaining educational television (ETV) service from the common carrier_for service between Urbana and Chicago made it necessary for CERL to develop special (and more expensive) modulator/demodulator equipment capable of multiplexing up to four consoles on each voice-grade telephone line. This equipment should be adequate for the field-tests, but the change introduced delays which still persist, and it is unfortunate not to be able to field-test PLATO using ETV service at this time. *Recent reports from CERL show that PLATO is available approximately 96% of the prime time (8 a.m. to 10 p.m., Monday through Friday), with roughly 7 hours between failures, each averaging 6 minutes in duration. The probability of a class (with scheduled access to consoles) being interrupted by a system failure is therefore about .1. These system problems have had their effect upon project cost and schedule, not to mention staff effort at CERL. They have also tried the patience of a number of remote PLATO users, who understandably expected the service to be commensurate with the cost (presently estimated to be roughly $10,000 of capital investment per console). These system problems, however, seem to be of a transient nature, inherent in any new system. Furthermore, some of the solutions to these problems are clear improvements in the PLATO design.** One problem, however, could prove to be more lasting, and therefore of greater significance in the long run. The problem referred to is the difficulty in preparing high quality courseware. It is hoped that a good deal of the delay thus far is developmental, due to the need to recruit and train a staff, and to discover how to use PLATO's features to good advantage. lf this theory proves optimistic, and if improved training, authoring aids, techniques, and standards don't increase productivity substantially, we may discover that the original estimate of about 40 man-hours of effort per student-contact-hour of courseware is off by a large factor, at least for authoring teams developing courseware specifically for use by other classes and faculty. The unexpected increase in the cost to develop courseware, together with the system problems summarized above, have increased the cost of the project by roughly 50%, and have forced a full year's delay in the commencement of the fieldtests. Use of PLATO this fall will be limited to fewer students and classes, and less courseware than planned even six months ago. Furthermore, the task of evaluation, already complicated by a wide range of content and grade levels at widely scattered locations, is made even more difficult by the delays and changes in plans. For example, the interpretation of the baseline and other evaluation data will be less straightforward, and less data will be available concerning PLATO use, owing to the delay. The National Science Foundation has reviewed the PLATO project and plans during the past 9 months. Although the problems noted above are not insignificant, they do not obscure the promise of this system, nor the accomplishments toward mounting a major field-test of it. The problems that have occurred, especially those requiring engineering solutions, have been addressed by CERL in a promising manner. A field-test of PLATO seems even more important than it appeared to be when planned three years ago. The Foundation has therefore agreed to share with the University of Illinois the additional support required. CERL is proceeding with plans to begin a two-year field-test in September of 1974, one year later than originally scheduled. **For example, CERL has designed and built an electronic (fast access) memory subsystem, that can be added to PLATO in such small quantities and at such a low cost as to provide every console with immediate access to its own lesson, thereby eliminating the dependence upon scheduled classes.