The concept of action-research has been developed during the last decade, mainly at the Research Center for Group Dynamics, University of Michigan, Ann Arbor and at the Commission for Community Interrelations of the American Jewish Congress—centers founded by the late Kurt Lewin whose original and creative mind has made many contributions to social-psychological and sociological research. I owe my acquaintance with this new approach to the Research Center, particularly to Ronald Lippitt and Alvin Zander. Yet most of the following observations are based on my research experience during the last four years. The responsibility for what I am saying is, therefore, completely my own.

The developments in 20th century physics which have brought into question the status of causality in subatomic phenomena are common knowledge today in the philosophical world. For the purposes of our discussion attention focusses on the quantum-mechanical solutions which describe atomic states by probability distributions instead of by exact values of observable variables.

This paper considers combinations of frictionless mechanical amplifiers arranged in a cycle so that the amplified signal passes around through an endless chain of such systems the output of the last system being fed into the first. It appears that while each amplifier may be reversible in the absence of friction, the effect of the combination is such that the signal would be unable to pass around the cycle indefinitely, using the same energy over and over again. It is felt that the assumptions upon which this cycle is based are related to some of the conditions met with in consideration of open thermodynamic systems such as heat engines and living organisms when in finite environments.

In 1948 Hempel and Oppenheim proposed an explicatum for the concept of systematic power 1, pp. 164–167. Since that time some shortcomings have been found in this first attempt. It is the purpose of this paper to show that one can keep the basic approach of the ‘48 paper, and overcome the known disadvantages by means of changes in the details of the definition. In this improvement certain tools will be used that were not available in 1948.

In “Studies In The Logic Of Explanation” (Philosophy of Science, XV, 1948) Hempel and Oppenheim analyze the basic pattern of scientific explanation. One of the difficult problems which they acknowledge is “whether and how the analysis of explanation can be extended from the case where all general explanatory principles invoked are of a strictly universal or ‘deterministic’ form to the case where explanatory reference is made to statistical hypotheses.” It is hoped that the remarks which follow may contribute a little toward clarifying the issue.

The notion of synonymy has recently been severely criticized, and its replacement by graded, continuous notions of one or another sort urged on general grounds. At the same time, it has usually been assumed both by critics and defenders of the notion, that synonymy and indirect discourse are in the same boat, that analyzing the latter, for instance, requires no more than an acceptable decision on the former while it requires at least that. Defenders of synonymy have thus thought it sufficient to apply their interpretations of this idea to indirect discourse, while opponents have not thought to attack such application save by way of an attack on synonymy.

Publication in this journal recently of a paper by A. Grünbaum has induced me to submit a treatment of the same subject. Like Grünbaum, I shall use the framework of special relativity, but my analysis will be different. Some of his arguments, I believe, are inadmissable. I shall pose the problem in a form different from his, but the solution to be described will be applicable to his case as well.