(26) 6. Bibliographic Data Entry A number of our users have developed program6 for automation of preprint and bibliographic files. These vi11 be reviewed and a suitable program adapted for public use. 7. Time-Oriented Medical Records Support and Other Patient Data Base studies By the fall of 1972, we hope to review all the work per- formed to date on tine-oriented medical records in order to select the optimal file desiga, the data entry foraat, and retrievable wocabulary for the near term usem. Our goal will be to provide the system tools and the systems or applications programming that will yield generalized programs which can be used by many different departments in the fledical Center, Ue have found that impleaenting the system in the Oncology Depart- ment required some changes froa the programs used for the Immunology Clinic. Ue have been impressed by physician acceptance of the time-oriented medical record. Please refer to the user project descriptions for Drs. Pkies and Slilliaa for additional explanation of the time-oriented medical record concept. Additional patient data base studies will be undertaken with the follouing goals: a. b. C. d. Obtain as such information as possible concerning new file systems to be supported by IB8 and other aanufac- turers. Review the Stanford SPIBBS spstea as to its applicability. Honitor carefully the problems encountered at Stanford ia use of the time-oriented medical records. Through literature, search, and travel remain well informed of work being performed elsewhere. Assign at least one Computer Science department graduate student (research assistant) to the problem of file structure design for patient medical records. The goal of these studies vi11 be to provide a broad base of information on file system technology and patient data base systems that vi11 be used to guide Stanfordas development and implementation of medical information systems, (27) C. Harduare Acusitions w---w-- -1a-1 Rental and purchase funds are included in the FY73 budget for several additional items of hardware. A discussion of each itea follows, 1. Terminals Several 2741 terminals will be released and funds will be used to acquire tvo additional high speed hard copy terminals and four CBT terminals. Ye are especially interested in terminals capable of operating at 1200 baud. The choice vi11 be limited to standard EBCD, Correspondence Code, and ASA terminals, if at all possible. 2, Public Printer/Plotter The availability of a relatively fast printer/plotter device for general use has been requested by our users. We have been looking at several electrostatic printers, as such a device provides extremely fast and flexible printing and plotting. Hovever, electrostatic devices place a high demand on CPU and channel resources. A further review of available plotters vi11 be conducted before a decision is reached. 3. PDP- 11 Equipment An item is included for purchase of additional core memory for both the development PDP-11 and the front-end PDP-11. Several additional devices (printer/plotter, DEC tape, voice drum) are to be attached to the front-end PDP-11. Available core will not be adequate to service these devices. The current 8k of core on the development PDP-11 is minimal for DEC's operating systen. Purtherrore, several users have been unable to perform some operations due to the core limitation. An additional 8k is required. 4. Ampex Disk Drive Current 360 disk utilization approaches full capacity. Addition of a new user pack may be necessary unless significant savings result from the new file compaction algorithm. (28) The disk configuration is eight on one control unit, eight ylus spare on the other. On the first control unit, the eighth drive serves as a spare since only fifteen disk packs must be permanently mounted. If we decide to add an additional user pack (always permanently mounted), rental of a second spare drive guarantees better system reliability. I- 3. Riscellaneous This category includes interfaces and' cabling for the voice drum and additional acoustic couplers and modems (part- icularly for 1200 baud terminals). D. Listen Softuare -w--Y- The specific suggestions for system softuate tasks are derived from four criteria: 1. The software should be stabilized as soon as possible. 2. System documentation must be completed. 3. Additional or improved user services should be aimed toward attracting additional income. 4, all changes in the form of additional features should be disallowed and the system frozen by the end of the grant year. Three major modifications (points 5, 6, and 7) are in direct conflict with rguirenents one and four. tiowever, they dre also the prinarp tasks for achieving requirement three. 1. System Stabilization and Cleanup The effort here vi11 concentrate on the elimination of known inconsistencies and errors in the system. Yherever easily possible, attempts will be made to reduce the size of the software. 2. Dot unen ta tion A considerable docnseatation effort is planned, The aims are to remove incompleteness and redundancies in the current documentation of the system software. A general paper should be written explaining the system design, the general functional flow, and the interactions of the various components. Also, a comprehensive table of contents and a cross index to each subject category is useful. (29) Cleanup of documentation and the softuare will benefit any transfer of the system code to another machine and it will expedit the maintenance of the system by a reduced staff. 3. Support of Additional Hardware We anticipate attaching a number of new peripheral devices to the system. A list of possibilities is discussed in Section C above. Additional software will be required in the front-end PDP-11 system for support of the voice drum, this year's grant), DEC tape (purchased from and a printer/plotter. Depending upon the type of printer/plotter selected, we may be able to adapt manufacturer-supplied software. With the installation of the Hemorex 1270 controller, session-controlling devices (terminals) and graphics devices are often synonymous. A few of the planned projects are: a. b. C. d. Where necessary, add specialized terminal support. An exanple is the inclusion of the Correspondence Code translation table as developed by the Stanford 360/67 facility for the unique features of the IBB Kagnetic Card Selectric Typewriter terminal (BCST), The new TBKTEOHIX terminals (4010), if selected, ized support. will require some special- Add a few terminal control commands, such as SET UPPEB/ UPLOW and SHOW CASE for specifying upper and lover case options. or TTY), Other commands determine the terminal type (ASA and suppress or allov transmission of the ACHE light box characters. For high-speed displays, incorporate modifications to the LIST command so that output is suspended every I4 lines. This allows the user to scan a page of lines and use BREAK if he is interested in stopping at that point. Implement page mode output for displays. A full screen of data is displayed directly from the user's in-core array or variables with a single command. The user may then modify the contents of the screen and read it back with another command. Some of the problems in implemen- tation of this capability are identical to those discussed under 'Extended Logon'. (30) 4. Additional Satellite Hachine Support Satellite machine support will focus on tasks essential for a smooth transition to SUIiEX and other long range programs expected to succeed ACHE. Ia addition, assistance vi11 be pro- vided to small machine users wishing to connect to the neu small machine multiplexor. 5. Batch and OS Services The provision for a user batch service seems necessary for several reasons: Current methods for submission of user disk-to-tape jobs and for submission of LISP overnight work are inadequate and operationally difficult to control. Our present method for providing routfne printing and punching services invites some danger to the reliability of the ACME system. Occasional long delays produce poor turn-around time. A number of jobs now routinely executed fro@ an AC&lB terminal need not be run interactively. Scheduling these jobs for overnight batch frees both the terminal and the CPU for additional on-line work. Recent hardware monitor tests indicate that the CPU is busy for 90%-95X of available cycles during peak daytime hours. Also, there is little free core available during the same hours. Consequently, only a small printing/punching utility region could be supported during the prime shift. Bowever, sufficient core and cycles are available during evening and overnight shifts to accommodate a limited load of user batch work. Provision for a batch service requires the completion of these tasks: a. An OS-type task which will convert ACfM-type files to OS files and the converse. The design document on this feature has been completed. Both direct access and tape data sets vi11 be supported. b, Incorporation of the HASP system into the OS supervisor. HASP facilities are preferred over conventional OS services for both better operational contlcol and for its smaller core and disk requirement. (31) C. kevieu of the WYLBUB/EASP interface and its adaptation to the ACHE system, if possible. Otherwise, the developaent of a capability to communicate uith the HASP in-core reader. d. Development of a small library of standard OS procedures for routine utility operations (printing, punching, etc.) e. Adaptation of HASP accounting. Development of a program to merge HASP accounting with time-sharing accounting. 6. PL/ACHE Object nodules The ability to save compiled code on a PL/ACHP program and reload it at a later time has been repeatedly requested, Unfortunately, we have avoided the subject primarily because of the lack of a definitive statement as to its impact on the current design of compiled code. Reloading of compiled code saves user setup time. It also frees CPU cycles now used for recompilation for more productive purposes. Due to the time constraints of the next year and due to the uncertainties mentioned above, this project is initially planned as a two man-month study and design specification. Actual isplenentation plans will be deferred until the level of effort is known and until the choice of a computer system is determined. 7. Extended Logon As the ACClE system is currently configured, a user session can be controlled only by a device routed through the Flenorex 1270. Beferences to the desirability of expansion to logon services are to be found throughout this report. Specifically, some of the extensions are: a. Allow a satellite computer to function as a terminal, invoking programs, etc. b. Allow our front-end PDP-11 to act as a raster terminal, 'logging on' users from devices not necessarily capable of sending and receiving lengthy transactions (such as touch-tone telephones). 02) C. Provide a master/slave terminal concept, so that one copy of a user program can service multiple requests from multiple, geographically separate, users. d. Page mode support for CRT's requires extension to the 'one-line-at-a-tiue' node of operation. The precise technique for implementation has not been determined. However, a reorganization of high-level system control functions is absolutely necessary, Bxpansions to the GET/PUT and READ/HRITB stateaents of PL/ACHB are a possibly simple scheme for user commnications, We feel that this is the most valuable of all suggestions for additions to the ACHE system. The considerable design effort and its impact upon the total structure of the system software are more than justifiable. 8. Miscellaneous This category contains a number of small projects, none of which may produce any signif icant increase in system usage, but all of which provide gratifying additions to the current level of service. Compared to the level of effort anticipated for implementation of other proposals in this section, these projects require very minor * anpouer expenditures. a. A method whereby a user may determine his aggregate monthly charges from his terminal (SHOU CHARGBS cormand). b, Extensions to the Text Editor to allow relocation or duplication of lines of text within a data set (HOVE/ COPP commands). c. Addition of the UNPEOTECT command to corplement the PROTECT command. This command is necessary if a user vishes to delete or maintain an existing protected file. d. A technique is needed whereby a user can guarantee sufficient disk file space for writing data to his data set. A simple implementation is to provide an option wherein the user specifies the nuaber of disk blocks required at OPE# time. The blocks are held in reserve for him (assuming they are available) until either they are consumed or the file is closed. (33) v. AD~IIISTRATIBB OBCAPXZATIO1 Stanford University has an Associate Provost for Computing anl reports directly to the Provost and Academic vice President. The Provost for Computing and heads the operation of three major facilities: Campus facility, ACBE facility, and SLAC computing racility. The current SCC director, Charles Dickens, has made a number of organizational and procedural changes during the past year in an effort to find an optimal organization for Stanford's system programming efforts. One finding uas that organizational rather than functional lines seen to be the most appropriate in the Stanford setting. For six months ue had a single systems group manager for all three facilities - SLAC, Campus, and ACBE. Soae benefits of the trial period were the creation of a better reporting system, cross-fertilization of ideas. and much more interaction among the systems programmers. Some of the dis- advantages were a longer communications chain, travel time between facilities, and some loss of control by facility directors. The operations managers of the three facilities mere also brought together in one functional arm of the organieation. Cross- facility training among operations was partially completed. Reporting schemes were unified, and an exchange of ideas nas accomplished. Today, the functional organizations exist for purposes of communication and coordination. Pormal lines of authority, houever, have been returned to individual facility directors. We have had a number of meetings in the past year with the Hospital Data Processing facility director and his staff. A new era of cooperation and communication between Hospital Data Processing and ACME has arrived. The personnel currently on the ACBE payroll, their job func- tion, and percent of full time equivalents are listed below: (34) CURRENT ACAE PERSONNEL NAflE Jamtgaard, R Wiederhold, G Rindfleisch, T. Hundley, L. Prey, R. Granieri, C. Len, Y, Riller, S. Briggs, R. Stainton, R. Bassett, H. Crouse, L. Whitner, J. Aranda, H. Baxter, E, Class, C. Cower, R. Sutter, J. Hatous, J. Rieman, J. Cannon, D. XPTE 100 Director 20 Consultant 50 Systems Analyst 100 Systems Programmer 100 Systems Programmer 100 Systems Programreer 100 Systems Programmer 100 Systeas Programmer 20 Systems Programmer 100 Systems Programmer 100 Scientific Programmer 100 Scientific Programmer 100 Scientific Programmer 100 Secretary 100 Administrative Asst, 100 Operations Manager 100 Computer Operator 80 Computer Operator 100 Computer Operator 60 Computer Operator 50 Dispatcher JOB TITLE Total FTE 17.9 (35) VI. BUDGET A. Resource Expenditures SUMMARY Total Resource Expenditures 1. 2. 3. 4. 5. 6. 7. a. 9. 10. 11. Actual Previous Budget Period Personnel: a. Salaries & Wages b. Staff Benefits Subtotal 239,329 32,851 272,180 Consultant Services 912 Equipment a. Main Resource-Rented b. Main Resource-Purchased C. Supporting Equipment d. Equipment Maintenance Subtotal 384,542 40,848 1,951 6,145 433,486 Supplies 15,873 Travel 3,047 Engineering Services 11,818 Publication Costs 3,031 Other a. Computer Services (l) b. Other Subtotal 8,272 9,531 17,803 Subtotal - Direct Costs 758,150 Indirect Costs 141,205 Total Costs 899,355 (1) Includes education courses (2) Assumes $443,976 exempt equipment costs and user income of $360,000. Current Budget Period Estimated Next Budget Period 248,690 277,162 37,505 44,231 286,195 321,393 1,000 1,000 389,458 390,596 54,479 50,000 1,260 1,260 14,500 18,805 459,697 460,661 9,200 9,700 4,000 4,000 30,000 31,500 2,000 2,500 3,100 7,900 10,100 11,200 13,200 19,100 805,292 849,854 76,600 21,771 881,892 871,625 (2) (36) B. Justification for FY1973 Budget Fiscal year 1973 will be the final year for the ACME Grant. The gross operating costs will be slightly higher than fiscal year 1972, primarily due to salary increases, some incremental equipment rent and maintenance, and some additional programming help for users. The emphasis in this final year will be one of pulling loose ends together and preparing for a transition to alternate hardware for research and service computing. In addition, strong effort will be made to improve the income posture of the facility so that it can survive in future years without the financial guarantees provided under the ACME Grant. The budget for fiscal year 1973 assumes that the staff level achieved in fiscal year 1972 will be retained. The facility is currently recruiting a replacement for Gio Wiederhold who has had primary responsibility for liaison with users and supervision of the consultants. Gio Wiederhold assumed new responsibilities in the Hospital Data Processing Facility in March, 1972. ACME continues to pay a fraction of his salary for consulting on system problems and planning. An average salary increase of 5% has been included in the budget for fiscal year 1973. Equipment rental costs have been budgeted at the current rental rates. The overall equipment rental budget is slightly higher than fiscal year 1972 due to the installation of Ampex core. Similarly, the hardware maintenance budget is up slightly due to acquisition of terminals and aging of interface hardware built during the past six years, and maintenance of the small machine equipment pool acquired over the past 18 months. Some miscellaneous budget elements include funds for training of staff on the PDP-10, acquisition of some terminal service on PDP-10's run by other institutions, and some additional communications costs associated with the move to multiple speed terminal support made in April, 1972. Capital equipment requested for the new budget period is as follows: 1) Four CRT terminals 2) Electrostatic Printer/Plotter, or expansion of smsLL1 machine equipment pool u.,600 22,000 3) 8K of core for development PDP-11 5,700 4) 4K of core for production PDP-11 3,300 5) Two 30 character per second typewriter terminals 5,000 6) Acoustic couplers, modems, etc. 1,400 TOTAL $50,000 (37) The CRT's and 30 character per second typewriter terminal will replace IBM 2741 terminals which have been rented by the facility in the past. The terminals are less expensive and will provide faster output and a savings of programmer time. The printer/plotter to be acquired has not been selected yet. This item is incorporated in the budget because graphic support continues to attract more users. At the present time the ACME facility has no plotter to support its users. Three user groups have plotters attached to ACME and they find the provision of service to miscellaneous users more than a trivial nuisance. Provision of good interactive graphics support on the ACME system will demand that the facility offer plotting to its users. Another problem which we have had is the slow turnaround on print jobs caused by the speed of our printer and the system degradation caused by use of the printer while PL/ACME is running. This latter problem is induced by IBM hardware and software constraints which cannot be fixed without a very sizable expen- diture of effort. The core for the Development PDP-11 and Production PDP-11 systems will permit use of the DEC Fortran compiler on the development system and will remove the limitation which has hampered the efforts of several development groups. The added core for the production PDP-11 will be needed only if several additional user devices or services are inter- faced through the PDP-11. We have barely enough core to mount the voice drum at the present time. We assume that additional special devices will be needed. Funds for acoustic couplers modems, etc., will be needed to effectively use the new terminals from remote locations. Also, a number of medical school groups are being moved to space adjacent to but outside of the current Medical Center buildings. For such users, telephone service for terminals is a must. This requires expansion of the communication equipment available. Users pay for their own terminals and acoustic couplers but the facility helps to respond to their temporary require- ments or new requirements pending delivery of new equipment. (38) C. Kxpenditure Details DIRECT COSTS ONLY 1. Personnel Director's Office Systems Analysis Systems Programmers Applications Programmers Research Assistants Operations Secretarial & Administrative Subtotal, Salaries Staff Benefits TOTAL PERSONNEL 2. Consultant Services August 1, 1971- August 1, 1972- July 31, 1972 July 31, 1973 32,929 23,582 4,633 10,600 94,306 97,960 39,042 63,470 9,852 7,150 52,430 58,680 15,498 15,720 248,690 37,505 277,162 44,231 2% ,195 1,000 321,393 1,000 09) C. Expenditure Details DIRECT COSTS ONLY 3. Equipment Major Equipment 1052 1403 2050 2314 #l 2314 #2 2361 2401 2403 2540 2701 2702 2821 Subtotal August 1, 1971- August 1, 1972- July 31, 1972 July 31, 1973 Console Typewriter Printer 600 LPM Additionsl CPU (F) Dir. Acess Storage Dir. Acess Storage Core Storage Mag. Tape Unit Mag.Tape Unit Control Card Reader Punch Data Adapter Unit Transmission Control Control Unit Ampex DC 314 Ampex DC 314 Ampex ECM-50 Memorex 1270 Disk Packs (IBM 2316/3) (25) Terminals (2741) (11) IBM 1800 add. units 1442 1826 1856 Subtotal, 1800 Unit Record 029 1,192 1,259 TOTAL, RENTALS 382,718 380,373 635 8,397 92: E 19 h8 27,291 3,377 8,971 6,947 10,561 16,834 10,937 20,068 22,243 77,949 -mm-- 635 8,wi 101,817 -w--m -s--m 355,619 2,114 11,763 --mm- 3,377 8,971 7,157 10,728 --M-m 10,937 36,304 36,304 119,922 16,248 360,842 (25) 2,100 (4) 4,109 2,638 2,671 7,691 7,691 1,701 1,701 12,030 12,063 (40) C. Expenditure Details DIRECT COSTS ONLY Purchased Equipment August 1, lg`i'l- August 1, 1972- July 31, 1972 July 31, 1973 RPQ's for Beehive Terminals 381 PDP-11 System 38,466 DEC Dual Tape System 8,700 1200 Baud Modems (2+ Power Supply) 750 Computer Terminals (2) 6,182 Four CRT Terminals Electrostatic Printer/Plotter, or expansion of small machine equipment pool 8K of core for development PDP-11 4K of core for production PDP-11 Two 30 character per second typewriter terminals Acoustic couplers, modems, etc. 54,479 Data Set and Line Rentals 8,000 Maintenance (Under outside contract)14,500 12,600 22,000 5,700 3,300 5,000 1,400 50,000 a ,400 14,805 Total Equipment 405,218 403,578 (41) C. Expenditure Details DIRECT COSTS ONLY 4. Consumable Supplies Office Computer Subtotal, Consumable Supplies August 1, 1971- August 1, 1972- July 31, 1972 July 31, 1973 3,700 3,700 5,500 6,000 9,200 9,700 5. Travel Frey -U. of Missouri Comp. 8/G/71 -h%&, New York 8/9-12 -Automated Health Sys., Wakefield, Mass., 8/13 Nozeki -WESCON, S.F., 8/25 Wiederhold-SIGPLAN Symp., Purdue Univ., 10/24-26 Jamtgaard -MUMPS Sys., Boston Gen. Hosp., 11/28-30 Jamtgaard -Conference, S.F., l/19- 1/21/72 Wiederhold-Conference, S.F., l/19-21 Wiederhold-Conference, Washington, D.C., l/28-29 Grsnieri -SHARE, S.F., 3/6-10 Wiederhold-SHARE, S.F., 3/6-10 Granieri -IBM Class, S.F., b/12-14 Class -SJCC, Atlantic City & DEC, Boston 5/15-18 777 9 359 426 78 78 364 ;t 33 525 All Other Travel 1,;57 Subtotal, Travel 4,000 6. Engineering Services 30,000 7. Publication Costs 2,000 a. Computer Services 360/67 2,500 PDP-10 and Line Charges ---- Staff Training 600 Subtotal, Computer Services 3,100 4,000 31,500 2,500 2,500 3,000 2,400 7,900 (42) C. Expenditure Details DIRECT COSTS ONLY 9. Other Expenditures Audio-Visual Presentation Books and Periodicals Postage and Freight Telephone Physical Plant Technical Services Subtotal, Other August 1, 1971- August 1, 1972- Lull 31, 1972 July 31, 1973 ---- 200 2,000 7,000 100 800 10,100 700 200 2,000 7,000 300 1,000 11,200 GRAND TOTAL -- DIRECT COSTS 805,292 849,854 (43) D. Summary of Resource Funding BUDGET PERIODS Source of Funds Computer Equipment - Service Biotech. Resources Branch Support Amount of Current Award: Line (5) of Award Statement Adjustment from Prior Periods: 1. Line (4) of Current Award 2. Balance of 270 x/y Proceeds Total BR Support TOTAL FUNDING Actual Previous Budget Period 170,596 675,747 76,459 ~27,275' 724,931 895,527 Current Budget Period 270,000 573,755 27,275 5,292 606,322 876,322 Estimated Next Budget Period 360,000 511,625 511,625 871,625 (44) ooLAoooooooo wwcxlooooo~oo . . . . . . . . . . . CUQD~~--C--~~CU~~~ r-luhi~"~~~ mcu n m oou-lLnooooooo cucucdmcucucucudcuN 00000000000 00000000000 . . . . . . . . . . . mmmonoontnotn wgtg-g;gdgmt-a wo n et n nn "I= r-i r-ir-l 000 woo . . c;mn 2;`L"r-l 000 cucurl (46) z 000000000000000 000000000000000 . . . . . . . . . . . . . . . . ~-olnmr-lmmcuch\mcucvar-u2 OdOt-lnm~Lnt?Omal~mLowf ~w~~cua3~oomcr)mdm~r--3 c.7 n 0% n rf rf aj 5 m 2 (U 2 LT r-" cf aT c-l m (47) F. Income Projection: A revised rate structure was implemented effective April 16, 1972. In general, rates for use of ACME were increased by 25% to 30%. Some users computing bills were doubled as a result of the new increase. Actual income for the month ending May 15, 1972 amounted to $31,500. Some users will reduce their utilization due to the higher rates. A conservative estimate of income for fiscal year 1973 is $360,000. This figure does not include the terminal service fees which cover terminal rent plus other services to the ACME user community which could not be covered by the current funding level of the ACME grant. Some major users of the current system will be transferring loads to dedicated small computer systems in the next 12-month period. One such user is Dr. Stanley Cohen whose Drug Interaction Program should be running on a dedicated PDP-11 by early 1973. Another such user is Dr. Howard Sussman in the Clinical Laboratory; programs are being written now for use of the newly acquired Sigma 3 by the Clinical Laboratory. Both Drs. Cohen and Sussman will continue to make some use of ACME for data analysis but the income from these two sources will be much reduced over current levels. On the next page you will see a table which shows the monthly income collected by ACME for the past nine months. Based upon this experience and the estimated impact of the higher rates, it is felt that the $360,000 is a reasonable estimate of next fiscal year's income from computer service fees. (48) f (49) VII. PROJECT DESCRIPTIONS CORE RESEARCH PROJECT DESCRIPTIONS Staff Core research and development projects carried on by the ACME staff during Fiscal 1972 have already been discussed in previous sections of this report. Principal Investigator Joshua Lederberg maintains one account for the purpose of system development and testing and another as a utility file for demonstration programs- These projects, PILOT1 and PILOT2, are listed in Category 6 of the Summary of Computer Resource Usage. Users During portions of Fiscal 1972, two user projects had core research status: DENDRAL and the Drug Interaction Project. Since both began to be charged for their usage during Fiscal 1972, they are listed below as a combination of core and collaborative research, CORE AND COLLABORATIVE RESEARCH PROJECT DESCRIPTIONS Investigator: Edward Feigenbaum, Joshua Lederberg, and Carl Djerassi. Dept. of Chemistry, Computer Science, and Genetics Project Began 1970 Project: DENDRAL Realtime Chargeable use began May 1971 The DENDRAL project involves collaboration between the Instrumentation Re- search Laboratory operating under NASA grant NGR-05-020-004, investigators operating under NIH grant RR 00612-02, and ACME. The emphasis of the DENDRAL-ACME efforts is computer science, while that of IRL-ACME endeavors is data acquisition and computer-instrument control. The DENDRAL project aims at emulating in a computer program the inductive behavior of the scientist in an important but sharply limited area of science: organic chemistry. Most of the work is addressed to the following problem: given the data of the mass spectrum of an unknown compound, infer a workable number of plausible solutions, that is, a small list of candidate molecular structures. In order to complete the task, the DENDRAL program then deduces the mass spectrum predicted by the theory of mass spectrometry for each of the candidates and selects the most productive hypothesis, i.e., the struc- ture whose predicted spectrum most closely matches the data. (50) Core and Collaborative Research Project Descriptions (cont'd.) The project has designed, engineered, and demonstrated a computer program that manifests many aspects of human problem solving techniques. It also works faster than human intelligence in solving problems chosen from an appropriately limited domain of types of compounds, as illustrated in the cited publications. Some of the essential features of the DENDRAL program include: Conceptualizing organic chemistry in terms of topological graph theory, i.e., a general theory of ways of combining atoms. Embodying this approach in an exhaustive HYPOTHESIS GENERATOR. This is a program which is capable, in principle, of "imagining" every conceivable molecular structure. Organizing the GENERATOR so that it avoids duplication and irrele- vancy, and moves from structure to structure in an orderly and pre- dictable way. The key concept is that induction becomes a process of efficient selection from the domain of all possible structures. Heuristic search and evalua- tion are used to implement this "efficient selection." Most of the ingenuity in the program is devoted to heuristic modifications of the GENERATOR. Some of these modifications result in early pruning of unproductive or implausible branches of the search tree. Other modifica- tions require that the program consult the data for cues (pattern analysis) that can be used by the GENERATOR as a plan for a more effective order of priorities during hypothesis generation. The program incorporates a memory of solved sub-problems that can be consulted to look up a result rather than compute it over and over again. The program is aimed at facilitating the entry of new ideas by the chemist when discrepancies are perceived between the actual functioning of the program and his expectation of it. The DENDRAL research effort has continued to develop along several dimen- sions during Fiscal 1972. The mass spectra of some previously uninvestiga- ted compounds were recorded. The computer program has been extended to analyze the mass spectra of a more complex class of compounds, using new kinds of data. The artificial intelligence work on theory formation and program generality has also progressed. Many mass spectra were taken to gather more data for the DENDRAL program. The analysis of the mass spectra of a variety of functional groups provided general mass spectrometry rules for the computer program. The spectra of many estrogenic steroids were taken to elucidate the mass spectrometry of this class of steroids and to provide data for a problem area new to DEN- DRAL. The estrogenic steroid problem is new in several respects: (1) 3.n working with steroids, the program deals with much more com- plex molecules than ever before; (2) 'the computer program uses element maps from high resolution data to resolve ambiguities; and