AN INTRODUCTION TO SIMPLICITY RICHARD RUDNER Michigan State University The f our papers which f ollow this introductory essay were presented at the 1960 meetings of the A.A.A.S. in New York City. They were given at a session devoted to the topic of Simplicity of Scientif ic Theories, and spons- ored jointly by Section L and the Philosophy of Science Association. For good reasons, each of the papers appears in this issue essentially in the f orm in which it was of f ered at the session.' Each of the papers was written independently of the others-indeed, the occasion of the meeting was the f irst time any of the authors had access to all of the other papers. None of the essays, theref ore, evidences the luxury, af f orded the present introductory remarks, f or ref lection on the other papers of the session, or f or response to the lively discussion which f ollowed the reading of the papers, or f or second thoughts in general. In this light, and considering the breadth of interpretation which has traditionally marked the notion of Simplicity, the relevance which each of the papers has f or the others is a relatively f ortuitous but nonetheless happy circumstance f or which we must be gratef ul. I shall comment brief ly below on the relationships which I f ind noteworthy among the essays. I think, however, that it will be more appropriate to begin by f ocusing attention on the importance of contemporary treatments of Simplicity and on what appears to be a pivotal distinction between two of the relevant senses in which the term is currently being used. Whatever may be the case f or the serenity or unself consciousness with which practicing scientists go about the business of accepting or rejecting theories, it will surely not be denied that the problem of constructing an adequate philosophical rationale f or such practice remains in its perennial state of crisis. The recent past has witnessed monumental and illuminating attempts (such as those by Reichenbach and Carnap) to provide that rationale essentially in the f orm of a logic of induction. For the purposes of our present concern it is not necessary to rehearse considerations of the cogency of the objections 1 Owing to its length it was necessary f or Prof essor Bunge to actually read a somewhat com- pressed version of the paper he contributed. The complete version is published here. 109 110 RICHARD S. RUDNER that advocates of "objective" or "statistical" theories of induction have hurled against advocates of "logical theories of conf irmation" or, conversely, to consider the objections which advocates of the latter view have hurled against those of the f ormer. It is not even necessary to go into the arguments which partizans of neither of these points of view have leveled against both, nor into those which partizans of either have leveled against such "third" f orces as recent theories of "Subjective" probability. The unhappy f act is that in the matter of cogent objections against theories of inductive inf erence, the recent philosophical literature provides us with an embarrassment of riches. The reason that none of these considerations need detain us f or the present, however, is to be f ound in the f act that even if any of the types of programs f or inductive logic mentioned above could be brought to the successf ul consummation its proponents apparently envisage, we would still not have been provided with a complete or general basis f or choice among theories. There are weights other than that of evidential strength whose assessment is a necessary condition f or rational (i.e. scientif ically reliable) choice among hypotheses.2 One of these additional weights we may ref er to as the cost associated with the acceptability of any hypothesis; and philosophers and many scientists (e.g., some who are concerned with Decision Theory) have in recent years come to give the explication of this notion something like the attention it has always warranted. Whatever the importance or the poignancy of the problems which attend the explication of cost, however, our concern here is not with it but with still a third weight whose explication is also a necessary condition f or the achievement of an adequate theory of inductive inf erence: I ref er, of course, to simplicity. Now, allusions to simplicity in the literature of Science and of Philosophy are innumerable and immensely varied in intent and nuance; and bef ore any f ruitf ul consideration of the topic or its importance can be undertaken it is necessary to delimit to some extent the range of our attention. This can be accomplished by f itting, with a minimum of procrustean f erocity, all of the varied ref erences to simplicity which we are heir to under a relatively un- complicated classif icational schema.3 Uses of 'simplicity' tilen, may be classif ied either as Ontological (i.e., extra-linguistic) or Descriptional (i.e., linguistic). Sub-classif ications under these main rubrics are Subjective (i.e., psychological) and Objective (i.e., non-psychological). Moreover, under the rubric, Descriptional, it is also f ruitf ul to distinguish Notational and Logical (or Structural) as f urther subclassif ications. A f ew examples will be suf f icient 2 In making this claim I do not, of course, intend to minimize the importance of the attainment of an adequate measure of evidential strength f or hypotheses as a desideratum of Philosophy of Science. I have urged more f ully than is appropriate here my conclusions about the insuf f i- ciency of any measure of evidential strength in [23] and [24], and more recently in a paper, "The Reducibility of Types of Weights in the Acceptance of Scientif ic Theories: Evidence, Cost, and Formal Simplicity," read in Stanf ord at the 1960 Meetings of the International Congress f or Logic, Methodology and Philosophy of Science. 3 The one we are about to employ is suggested though not in precisely this f orm in Chapter 1 of Dr. Ackermann's searching thesis, [1]. AN INTRODUCTION TO SIMPLICITY 111 to provide the very rough degree of clarity which is all we presently require f or these six category terms. Consider, f irst, ontological simplicity. It is quite clear that a great many people who have employed 'simplicity' or its cognates have used the term to attribute certain characteristics to the Universe rather than to our descrip- tions of the Universe. For such individuals it has been the extra-linguistic universe, or some segment of it, which is said to exhibit or to f ail to exhibit some degree or other of simplicity. Moreover, these attributions of ontological simplicity may be classed as objective or subjective in accordance with whether their import was that the universe is (or is not) simple independently of how we perceive it or whether the (extra-linguistic) universe is (or is not) perceived by us to be simple. If the intent of the usage is that 'simplicity' is a predicate of the universe independently of the perception of it, we shall classif y that usage as objective-ontological; if 'simplicity' is taken as a predicate of our (extra-linguistic) responses to the (extra-linguistic) universe, we shall classif y it as subjective-ontological. Attributions of simplicity both to the universe and our extra-linguistic responses to it, of course, abound in the literature of science and metaphysics. Such attributions have always seemed f undamentally obscure to me. Yet, having reminded ourselves of their occurrence, we need linger no longer over them than is required to note that it is not ontological simplicity but descriptional simplicity which is the f ocus of interest of theories of inductive inf erence and particularly of the f our essays here being in- troduced.4 Turning f rom ontological to linguistic considerations, then, we may, under the category of Descriptional Simplicity, at once distinguish two kinds: Notational and Logical. Whether any general distinction between the Notational and the Logical simplicity of descriptions is ultimately tenable is, again, not in point here. In employing it, I wish merely to call attention to such dif f ering properties of descriptions as, say, their brevity in contrast with the degrees of the predicates they contain. An attribution of simplicity to a description on the basis of such a notational property of it as the number of (e.g., alphabetical) characters it contains, independently of anyone's psychological responses to such a property, will be classed as an instance of Objective-Notational simplicity. On the other hand, an attribution of simplicity to a description on the basis of the f amiliarity of the notation, or its elegance, or its convenience, or its ef f iciency f or manipulation, or any aesthetic quality it has, etc., will count as an instance of Subjective-Notational Simplicity. In the sense just indicated, then, neither objective nor subjective notational simplicity is our topic of concern in what f ollows. What is being attended to is the logical (or, alternatively, the f ormal) simplicity of descriptions-and especially those descriptions which constitute scientif ic theories. Moreover, 4 This remark should be qualif ied: the survey of various types of simplicity in Prof essor Bunge's essay doubtless encompasses more than what is ref erred to as "descriptional simplicity" above. 112 RICHARD S. RUDNER since our interest is not in how people psychologically respond to logical properties of theories, we may characterize our f ield of attention as Objective- Logical Simplicity. Hereaf ter in these comments ref erences to simplicity, unless otherwise qualif ied, are intended as ref erences to Objective-Logical Simplicity. Realization of the importance of considerations of simplicity f or the Philos- ophy of Science is a phenomenon of the relatively recent past. This is not altogether surprising in view of the f act that advances in Logic, upon application of which much of the signif icant work accomplished has depended, are themselves phenomena of this century. Despite the importance of achieving an adequate explication of the concept, sustained and signif icant work on its accomplishment has thus f ar been undertaken by only a relatively small circle of philosophers. In the quite recent past this circle has slowly widened as interest in the problem has come to be quickened or inspired under the impetus of the positive and detailed results achieved especially by Prof essor Goodman. In any case, however slowly launched, work by an increasing number of able men, is now under way and we can look f orward with hopef ul excitement to the solution of problems about simplicity which once appeared well nigh insuperable. Perhaps the importance of attaining an adequate explication of simplicity can best be indicated by pointing out some aspects of its connection with systematism. On this score, Goodman's opening remarks in a recent article are as illuminating and pithy as any which have been made on the topic: All scientif ic activity amounts to the invention of and the choice among systems of hypotheses. One of the primary considerations guiding this process is that of simplicity. Nothing could be much more mistaken than the traditional idea that we f irst seek a true system and then, f or the sake of elegance alone, seek a simple one. We are inevitably concerned with simplicity as soon as we are concerned with system at all; f or system is achieved just to the extent that the basic vocabulary and set of f irst principles used in dealing with the given subject matter are simplif ied. When simplicity of basis vanishes to zero-that is, when no term or principle is derived f rom any of the others-system also vanishes to zero. Systematization is the same thing as simplif ication of basis. Furthermore, in the choice among alternative systems, truth and simplicity are not always clearly distinguishable f actors. (p. 1064, [12]). System is no mere adornment of Science, it is its very heart. To say this is not merely to assert that it is not the business of Science to heap up unrelated, haphazard, disconnected bits of inf ormation, but to point out that it is an ideal of science to give an organized account of the universe-to connect, to f it together in logical relations the concepts and statements embodying whatever knowledge has been acquired. Such organization is, in f act, a necessary condition f or the accomplishment of two of Science's chief f unctions: explanation and prediction. The work that has been done, and the work currently being done so f ar as it is manif est, on objective-f ormal simplicity cannot plausibly be viewed AN INTRODUCTION TO SIMPLICITY 113 to have brought us to a complete and adequate explication of the concept. It has, f or its students, had the contrary ef f ect of bringing a clearer realization of precisely how its problems ramif y and of how much remains to be done. Nevertheless, if much remains to be done, what has already been done can f airly be said to have gotten the enterprise well under way and to give great promise of continued progress. And this is a towering accomplishment; f or, hopelessness in the f ace of the problem of clarif ying simplicity or (more or less disguised) f light f rom the problem have been the almost universal responses of the keenest intelligences which in the past have conf ronted it and its manif estations in the literature. Let us, even if brief ly and too simply, review this accomplishment. Each of two major avenues of approach, has been f requently ref erred to as a concern with f ormal simplicity. Yet, the two are f undamentally distinct and there is reason to regard with some misgiving (as I shall try to show below), this classif ication f or one of the approaches. One of these lines of inquiry, the newer of them, has led toward an explication of the notion of the measurement of the f ormal simplicity of conceptual (i.e. predicate) bases of descriptional systems or theories. The considerable positive results which have been f orthcoming are essentially the contribution of one man, Nelson Goodman; and they are, without doubt, results which f all under the category of objective-logical simplicity as it has been delineated above. The second line of inquiry, an older one, is associated with "the problem of curve f itting." The notion that in conf rontation with data, representable as a set of graph points, the scientist should "induce to" the simplest ("smoothest") curve or f unction descriptive of that data, is of course, not unhackneyed. Such an admonition obviously raises the problem of how the simplest f unction among the alternatives available is to be discriminated. What would be cogent criteria of simplicity here? In light of what has been said above our interest will not be engaged by attempts to provide answers which are essentially ontological or subjective in character. In connection with criteria of objective simplicity, f or the curve f itting problem over the past f ew decades, the work of three men, Harold Jef f reys, K. R. Popper, and John Kemeny, (see bibliography appended), is especially noteworthy. Their problem is so of ten ref erred to as work on "the problem of inductive simplicity" that I shall f or the present adopt this term though I believe it to be somewhat misleading. In discussing results accomplished by Goodman on the problem of simplicity of predicate bases and by the others mentioned on the problem of inductive simplicity I can, of course, here do no justice at all to the f ormidability of the specif ic and detailed problems they address, or the complexity and precision of their ef f orts, or the perspicacity of their conclusions. Accordingly, no such justice will be aspired to; I shall be content to convey only the most general impression of the nature of each of these lines of inquiry in the hope that it may lead those unf amiliar with the relevant literature to items of the appended bibliography. It might be thought that attempts such as Goodman's to clarif y the notion 114 RICHARD S. RUDNER of systemic simplicity would have an obvious f ocus on the simplicity properties of sets of postulates. Thus, a normal f irst impulse might be to say that of two otherwise equally adequate theories the one with the f ewer postulates was objectively the simpler. But little ref lection is needed to show both that this suggestion is unhelpf ul and also that its very lack of promise leads naturally to consideration of the simplicity of a theory's set of primitive predicates. For, the f inite number of postulates of any theory can be trivially reduced to 1 by the simple operation of conjunction. By the criterion of number of postulates every theory would be equivalent to some theory which was maximally simple. Nor would it be possible to ameliorate this unwelcome result by any evident stipulation regarding the number of conjuncts in a set of postulates. For if the import of such a stipulation is, f or example, that a postulate whose f orm is (1) f *gX is less simple than a postulate whose f orm is (2) hx then the def ectiveness of that stipulation becomes clear as soon as it is realized that it is always trivially possible to construct, i.e., to def ine or explicate a predicate, h, such that (3) hx=- (f - g$) will be logically true. Accordingly, any postulate of f initely many conjuncts is trivially reducible to a postulate of one conjunct and by such a criterion all postulates must be regarded as equally simple. Even this unelaborate example indicates that to get at a relevant sense of 'simplicity' we must go beyond considerations of the number of , or gross logical structure of , postulates and come to grips with the logical structure of the predicate bases of theories. Since it is plausible to assume that the theories we are interested in all share a common logical apparatus this means that attention turns to the f ormal simplicity of the extra-logical predicates. And this is, indeed, the route which Goodman f ollows. In the course of several years of work and through a process of increasingly successf ul modif ications he has been able to construct a calculus of predicate simplicity which provides a measure of the simplicity of predicate bases of every relevant logical kind.5 In general, and necessarily vaguely, Goodman's assignments of simplicity values may be thought of as depending on the manner in which the extra-logical predicates of a theory organize, by virtue of such of their logical properties as ref lexivity or symmetry, the entities comprising the total extension of the theory. In coming to understand the import of Goodman's work it is especially important to avoid a conf usion (not always avoided by earlier commentators on his work (see [1], [20], [25], and again especially [9]) between the simplicity of a basis and its power. The sets of predicates of two systems, S and S' 5 For an explanation of the crucial notion of relevant kind, see [12], and especially [9]. AN INTRODUCTION TO SIMPLICITY 115 are equally powerf ul if the sets are interdef inable. Suppose that no predicate of the set in S is def ined by any other in S. If the power of a basis were the same thing as its simplicity, "no simpler basis f or S . . . [could be arrived at than] . . . by taking all the predicates of S as primitive" (p. 430 [9]). But, it is precisely the greater simplicity of an S' whose primitive basis is "narrower" (i.e., whose basis systematizes through def ining the remainder of the predicates by a subset of the total number in the system,) over an S whose basis is the "widest" possible which we desire to measure. In the last analysis what we are af ter is the economy of a system; and just as we get an indication of the economy of an automobile not f rom its having gone a certain distance but f rom how much gasoline it requires to go that distance, so too with the economy of systems. The power of a system is strictly analogous to the distance driven of our car in that knowing it alone will not give us a measure of economy. To arrive at the economy of a system we require also some measure of the simplicity of its basis-and it is this that Goodman's calculus attempts to provide. What has been said above must here suf f ice as an indication of the sort of concern that the topic of simplicity of predicate bases involves. In connec- tion with inductive simplicity I shall not linger on any exposition since three of the f our panel papers which f ollow scrutinize the major work done on it. Perhaps, however, some f urther light may be thrown on its relationship to predicate simplicity and the propriety with which it has been placed in the category of Objective-Logical Simplicity, by the f ollowing considerations. First, as Dr. Ackermann points out in his paper (and in more complete detail in [1]), the concept of inductive simplicity as elaborated by Jef f reys, Popper, and Kemeny comes to depend on some such notion as the number of "f reely adjustable parameters" which occur in alternative hypotheses. Now, Ackermann's discussions have revealed very grave def ects in these specif ic treatments of inductive simplicity on, so to speak, their own grounds. But, the viability of some specif ic treatment is not what is at issue here. Even if such treatments were otherwise wholly successf ul, it is doubtf ul that they would f urnish any tenable criterion or test of simplicity-especially of f ormal simplicity. Thus, f or example, there seems to be no good reason f or believing that a hypothesis with n + 1 parametric expressions is less simple in the sense of Goodman's calculus than a hypothesis with n parametric expressions. Goodman's calculus of simplicity is, of course, not even applicable to hypo- theses. Moreover, the obvious suggestion to classif y one hypothesis as "simpler" than another if the sum of the complexity values of its predicates is less than that of the other is, on a little ref lection, seen to be of no avail. Apart f rom the f act that the sets of predicates, which are constituents of each of the alternative hypotheses to be assessed, will not in general be identif iable with sets of primitive predicates of theories, there are perhaps more decisive reasons f or the f ailure of the suggestion. For one thing, it will not in general be the case that the set of predicates f rom the hypothesis of n + I parameters, will have a higher complexity value than sets of predicates f rom alternative 116 RICHARD S. RUDNER hypotheses of less than n + I parameters. This will depend among other things, on whether or how some of the predicates in such sets are def inable by others in the same set. Thus, the criteria urged by proponents of inductive simplicity will yield judgments in conf lict with what might be thought of as the obvious application of Goodman's concept in the same situations. Of course, all that this indicates is that the two concepts6 are not identical; and this f act would probably not discomf ort the proponent of inductive simplicity. He would very likely maintain that he had always been aware his technique measured some other type of simplicity than simplicity of predicate bases-one whose assessment is, nevertheless, at least equally important. But if some other kind of simplicity, then what kind? Ontological simplicity surely is not at issue here; and if it were, could scarcely be def ended. Again, despite our initial characterization of inductive simplicity as f alling within the category of Formal Simplicity, it is puzzlingly dif f icult to make out just what f ormal or logical properties of hypotheses, i.e., statements, are involved. If con- siderations of gross logical structure (such as those discussed earlier in connection with postulational simplicity) are at issue, then we are at once driven to the f ormal structure of the constituent predicates. And here the logical characteristics, other than those indexical of simplicity in Goodman's sense, seem to have relevance to such measures as power rather than simplicity. On the other hand, if the relevant f ormal properties of hypotheses are construed as those having to do with their logical strength, then Goodman's and Barker's criticisms (and especially the counter-examples adduced by the f ormer,) in the papers which f ollow show decisively that it is a mistake both to identif y degree of logical strength of a hypothesis with simplicity and also to advocate the choice of the simplest hypotheses in the relevant situations. Goodman's arguments and especially his suggestion that simplicity of hypotheses is associated with their projectibility or the entrenchment of their predicates, are illuminating and stimulating. Yet, the suggestion, if cogent, seems to me only to show that in situations, to which some proponents of inductive "simplicity" have addressed themselves, projectibility, or predicate entrenchment, rather than the criteria adduced by those proponents will order our selection of hypotheses. What is not shown is that either projectibility or entrenchment are identical with, or even reliable indices of , simplicity in any tenable sense. To be sure, one might take the course of just identif ying entrenchment with simplicity in some special sense, but in the absence of any independent evidence f or supposing that degree of entrenchment is a f unction of the simplicity, in any plausible sense, of the predicates entrenched, this would seem to be merely a way of undesirably trivializing the entire problem. In point of f act, Goodman's explication of entrenchment (see [13], Chapter 4), gives us no reason to suppose any such connection. All of these considerations persuade me that whatever the cogent criteria may be f or ordering our selection of hypotheses in those situations (e.g., 6 I am assuming here that the proponents of inductive simplicity have "been af ter" some one concept-an assumption I shall shortly question. AN INTRODUCTION TO SIMPLICITY 117 curve f itting") that have hitherto been held to involve inductive simplicity,7 it would be misleading to classif y them as considerations of Objective-Logical Simplicity. One f inal category remains whose involvement might warrant the claim that objective measures of simplicity are being sought by proponents of inductive simplicity. We have not thus f ar discussed Objective-Notational Simplicity-but here too, such an obvio-us objective characteristic of a f ormula- tion as its length, seems whole uninteresting and unequal to the burden any portentous theory of inductive simplicity would make it bear. Patently, with the relevant qualif ications, a hypothesis with n parametric expressions is shorter than one with n + 1. Again, patently, it seems sensible to speak of such shorter hypotheses as being objectively notationally more simple than longer ones. But (as Goodman points out) any hypothesis is trivially reducible to one of minimal length, i.e., may be f ormulated in as brief notational compass as any other; so that on this criterion all hypotheses become in ef f ect objectively equally simple. And, in any case, perusal of the literature really precludes the belief that what the proponents of inductive simplicity have been af ter is an ordering of hypotheses with respect to their brevity. The f act is that those who held that there is a weight, properly called "simplicity," which must be assessed in "curve f itting" situations, seem initially to have been impelled to their analyses by recognition of the inf luence, not of objective characteristics of hypotheses, but of such subjective character- istics of descriptions as f amiliarity, or manipulability, or elegance, or in- telligibility, etc. No doubt, too, the psychological responses these terms signalize have, through the processes of "socialization" that entrants to the community of scientists undergo, come to be f airly standard among scientists; and perhaps this f act made the notion that there was some objective character- istic of descriptional simplicity operative in "curve f itting" seem more plausible than can actually be warranted. Whatever its etiology, however, such a conclusion seems especially misleading in view of the f act that Goodman has provided us with a quite distinct explication of objective simplicity. What I have been saying, then, can be summed up by pointing out that insof ar as the considerations which inf luence hypothesis acceptance in the "curve f itting" situation can properly be called considerations of simplicity they are subjective, while, on the other hand, insof ar as they are objective they are only misleadingly called considerations of simplicity. In what has gone bef ore I have already indicated that the papers of Good- man, Ackermann, and Barker, which f ollow are related through the f act that 7 I am assuming that these will be situations in which well articulated and independently conf irmed theories, which have as consequences hypotheses that "f it" the data, are not available. If such theories are available we have another kind of (broadly speaking,) inductive situation and considerations of say, the simplicity of the predicate bases of such theories may well be quite relevant. These latter kinds of cases, of course, are not in point here as is clearly revealed by the f act that not the f ormal properties of the curve f itting hypothesis but rather those of the theory f rom which it is derived become relevant. For an incisive discussion of this point see Ackermann, [1], especially the last section of Chapter II. 118 RICHARD S. RUDNER each critically addresses some proponent of inductive simplicity. They also appear to be united, it should be mentioned, in holding that simplicity is an important weight in scientif ic acceptances or rejections. In this respect, Bunge's paper serves as an admirable f oil f or the other three (and, accordingly, served to spark a lively discussion at the actual session) as well as to provide an extensive survey (one, moreover, valuably inf ormed by the thoroughness of his knowledge of physical science) of various possible interpretations of 'simplicity.' I do not f ind his arguments f or the conclusion, that Simplicity is not an important weight in the acceptability of theories, wholly compelling ones. Nevertheless, those arguments together with the erudition arrayed in their support stimulate and require the most serious consideration. For me the conclusions f ail to carry conviction on two counts. Though it would be inappropriate to argue these here, perhaps I can with propriety indicate what they are: First, Prof essor Bunge seems to arrive at his conclusion, concerning the insignif icance of the simplicity as a weight, on the basis of its f ailing to be a reliable sign of truth. But this seems to me to be an irrelevance, even if accurate. As indicated above, systematization seems to me as much a desidera- tum of science as is truth and nothing that Prof essor Bunge writes seems ponderably to assail the conclusion that simplicity is an important measure of systematization. Second, I f ind Prof essor Bunge's discussion sometimes vitiated by the opacity of some of his categories. In particular, I f ind myself still quite unclear about what he intends by "semantic," "epistemological" and "metaphysical" simplicity. It is doubtless inevitable, that the carrying out of so extensive a treatment of various types of simplicity in the short compass of an article, prohibits f ull elucidation of every category. These, however, seem so crucial to his discussion, and so interesting in their own right, that it is to be hoped that Prof essor Bunge will f ind the time to tell us more about them. REFERENCES [1] ACKERMANN, Robert. Simplicity and the Acceptability of Scientif ic Theories, Doctoral Dissertation, Michigan State University, 1960. [2] BARKER, S. F., Induction and Hypothesis, New York, 1957. [31 GOODMAN, Nelson, "Axiomatic Measurement of Simplicity," The J7ournal of Philosophy, LII (1955), pp. 709-722. [4] GOODMAN, Nelson, "An Improvement in the Theory of Simplicity." The Yournal of Symbolic Logic, XIV (1949), pp. 228-229. [5] GOODMAN, Nelson, "The Logical Simplicity of Predicates." The Journal of Symbolic Logic, XIV (1949), pp. 32-41. [6] GOODMAN, Nelson, "New Notes on Simplicity," The J7ournal of Symbolic Logic, XVII (1952), pp. 189-191. [7] GOODMAN, Nelson, "On the Length of Primitive Ideas," The Journal of Symbolic Logic, VIII (1943), p. 39. [8] GOODMAN, Nelson, "On the Simplicity of Ideas," The Yournal of Symbolic Logic, VIII (1943), pp. 107-121. AN INTRODUCTION TO SIMPLICITY 119 [9] GOODMAN, Nelson, "Recent Developments in the Theory of Simplicity," Philosophy and Phenomenological Research, XIX (1959), pp. 429-446. [10] GOODMAN, Nelson, "Sequences." The J3ournal of Symbolic Logic, VI (1941), pp. 150-153. [1] GOODMAN, Nelson, The Structure of Appearance, Cambridge, 1951. [12] GOODMAN, Nelson, "The Test of Simplicity," Science, CXXVIII (1958), pp. 1064-1069. [13] GOODMAN, Nelson, Fact, Fiction, and Forecast, Cambridge, 1955. [14] JEFFREYS, Harold and Dorothy WRINCH, "On Certain Fundamental Principles of Scientif ic Inquiry," Philosophical Magazine, XLII (1921), pp. 369-390. [15] JEFFREYs, Harold, Scientif ic Inf erence, Cambridge, 1957. [16] JEFFREYS, Harold, Theory of Probability, Oxf ord, 1948. [17] KEMENY, John G., "A Logical Measure Function," The J7ournal of Symbolic Logic, XVIII (1953), pp. 289-308. [18] KEMENY, John G. "A New Approach to Semantics-Part I." The J7ournal of Symbolic Logic, XXI (1956), pp. 1-27. [19] KEMENY, John G. A Philosopher Looks at Science, Princeton, 1959. [20] KEMENY, John G. "Two Measures of Simplicity," The Journal of Philosophy, LII (1955), pp. 722-733. [21] KEMENY, John G., "The Use of Simplicity in Induction," Philosophical Review, LVII (1953), pp. 391-408. [22] POPPER, Karl R. The Logic of Scientif ic Discovery, New York, 1959. [23] RUDNER, Richard S., "The Scientist Qua Scientist Makes Value Judgments." Philosphy of Science, XX (1953), pp. 1-6. [24] RUDNER, Richard S., "Value Judgments in the Acceptance of Theories." Philipp Frank, ed. The Validation of Scientif ic Theories, Boston, 1956. [25] SUPPES, Patrick, "Nelson Goodman on the Concept of Logical Simplicity." Philosophy of Science, XXIII (1956), pp. 153-159. [26] SVENONIus, Lars, "Def inability and Simplicity." The J7ournal of Symbolic Logic, XX (1955), pp. 235-250.