Unlike the days of Auguste Comte and his followers in the late 19th-Century, today there is a comfortable compromise between those interested in positive economics and those interested in the mathematics-based model-building techniques. Given this, it is not surprising that few economists today will be found waving the banner of ‘economic positivism’ or ‘positive economics’ or even ‘positive science’. Thus the absence of flag-waving does not mean that economic positivism is dead. Positive economics is now so pervasive that virtually all competing methodological views (except that of the most defeatist hard-core mathematical economists) have been eclipsed. The absence of methodology flag waving is thus easy to understand. There is no territory to dispute and thus no need to wave one’s flag.
The dominance of economic positivism is abundantly evident in current textbooks. Almost every introductory textbook explains the difference between ‘positive’ and ‘normative’ economics and tries to make it clear that economists are interested in positive economics and capable of fulfilling the demands of economic positivism. Why should economists be interested in positive economics? And has economics fulfilled the demands of economic positivism? These two questions will be my focus here.
Positive Economics vs What?
While every textbook clearly distinguishes ‘positive’ from ‘normative’ questions by characterizing the distinction with an ‘is/ought’ dichotomy, it is not clear that the history of the distinction supports such a dichotomy. John Neville Keynes (John Maynard’s father) is most often quoted to support the dichotomy despite the fact that he actually said, ‘The object of a positive science is the establishment of uniformities, of a normative science the determination of ideals, of an art the formulation of precepts‘ [1917, p. 22] – that is, he actually advocated a trichotomous classification. Unfortunately, the widespread reliance on the ‘is/ought’ dichotomy has nullified Neville Keynes’ best efforts to improve our understanding of positive economics.
While promoting ‘positive methodology’ in his famous 1953 essay, Milton Friedman tried to deny the ‘is/ought’ dichotomy by arguing that answers to ‘ought’ questions necessarily depend on a prior establishment of ‘what is’. Nevertheless, most critics of Friedman’s methodology think he was arguing against normative economics and thus assume that he was only arguing in favor of positive economics. The presumption seems to be that one must always choose between ‘is’ and ‘ought’ questions as if they were inherently mutually exclusive.
To be fair, there is a good reason to presume that ‘is’ and ‘ought’ questions are mutually exclusive. David Hume long ago argued that ‘ought’ statements cannot be deduced from ‘is’ statements and vice versa. The mere mention of ‘is’ and ‘ought’ in the definition of positive economics thus seems to demand a sharp dichotomy such as the one between positive and normative economics as defined in the textbooks.
In addition to the is/ought distinction, there are other dichotomies that seem to support the separation between positive and normative economics. There is the philosopher’s distinction between analytic and synthetic truths. There is the science-vs-art distinction which motivated early economic methodologists (such as Nassau Senior) – where ‘science’ was alleged to be about material truths, ‘art’ was considered to be about normative rules. More recent dichotomies are the objective/subjective, descriptive/prescriptive and rational/irrational, which are often considered direct correlates with the positive/normative distinction. And, of course, there is the more commonplace distinction between theoretical and applied economics that prevails in most Economics departments today.
To this list I wish to add one more distinction – namely, the romantic/classical distinction often found in discussions of nineteenth-century British literature. Specifically, I think one can recognize a distinction between ‘romantic’ and ‘classical’ postures concerning the realism of assumptions. While it might be considered romantic to assume the world is the way one would like it to be, it would be classical to dispassionately try to make one’s assumptions correspond to the way the world really is. For example, while a romantic egalitarian might wish that wealth be evenly distributed, a classical realist would contend that one should not assume distributional uniformities unless there are good empirical reasons to do so.
So, given all of these various dichotomies, how does one understand the nature of positive economics and why should one ever want to promote it? I think the reason why there are so many different distinctions raised in the discussion of positive economics is that each of them represents something that positive economics is claimed not to be. That is, most people understand positive economics more by what it is argued not to be than by what it is argued to be in fact. Briefly stated, we have only a negative understanding of economic positivism!
Positivism as Rhetoric
There is a sense in which the distinction between positive and normative is completely confused. Positive policy advisors are in effect always recommending that their policy is the best way to achieve the given ends. This is evident even in John Neville Keynes’ original discussion. It is difficult to conceive of a way one could ever avoid normative judgments. So, what is it that one is truly accomplishing when demanding that one’s economic research or advice conform to the dictates of positivism?
The idea of ‘positive’ economics is mostly a matter of rhetoric. The rhetorical purpose is also evident in the use of some of the other dichotomies. One can find books titled System of Synthetic Philosophy [Herbert Spencer 1896], Positive Philosophy [Comte 1855/1974], Scientific Management [Drury 1922/68], Objective Psychology of Music [Lundin 1967], Rational Economics [Jackson 1988], Descriptive Economics [Harbury 1981], and so on. Whenever an author is extolling the virtues of a theory by claiming it is a positive theory, he or she is usually asserting that it is not something of a scientifically unacceptable nature. What is acceptable in these matters is usually dictated by the prevailing view of ‘scientific method’. But it is not often clear why the term ‘positive’ must always indicate something acceptable or desirable.
In economics, the association between ‘positive’ and ‘descriptive’ seems to be a direct consequence of the reliance on David Hume’s view of the is/ought dichotomy. One describes ‘what is’ and prescribes ‘what ought to be’. The association between ‘positive’ and ‘applied’ economics and between ‘positive’ and ‘synthetic’ statements is rather confusing. While it is easy to claim that one’s theory is ‘positive’, it is more often thought that pure theory is not empirical and thus applied economics must be ‘positive’. So, what did Eugene Böhm-Bawerk mean by the title of his 1889 book Positive Theory of Capital? While it might be easy to see a connection between ‘positive’ and ‘synthetic’, their opposites do not seem connected. Hardly anyone would connect ‘normative’ with ‘analytical’ – except from the perspective that a normative conclusion is a logically contingent truth that depends on the acceptance of presumed values. But if analytical truths must be tautologies then, technically, the connection is rather weak. The post-war influence of the logical positivists and the retrospective influence of Max Weber have combined to make the rhetoric of positivism even more confused. Logical positivists were analytical philosophers who thought verifiable scientific knowledge is distinguishable from unverifiable ‘metaphysics’. The turn-of-the-century social scientist Max Weber is now credited as being a leader in developing the idea that scientific knowledge could be ‘value-free’. And, to confuse things still more, Karl Popper presented a critique of logical positivism based on the logical grounds that one’s theory makes a positive contribution to scientific knowledge only if it is falsifiable (most commentators seem to think means only that it is not a tautology). With all this confusion in mind, it may be difficult for us to determine even what ‘positive’ is not.
What everyone seems to think ‘positive’ is
Economic positivism as it is currently practiced seems to be available in four different flavors. The first and most optimistic version is what I will call Harvard positivism. It is represented by the recent attempts to develop ‘experimental’ economics and has its origins in the early teaching of Edward Chamberlin. At the other extreme is the weak minimalist version which I will call MIT positivism. Its weakness is due to the methodological view that says that to be of interest a theory need only be potentially refutable – there is no additional requirement that says it needs to be supported or tested by empirical evidence. In between these two extremes there are two more modest versions. One is what I will call LSE positivism, which does not require controlled experiments but does see economics as a scientific endeavor that emphasizes a necessary role for empirical, quantitative data. The other one is Chicago positivism, which includes both the simplistic instrumentalism of Friedman and the more complex confirmationism of Becker and Stigler.
Those positivists who advocate ‘experimental economics’ still comprise a very small segment of mainstream economics. The current movement seems to have its origin in the experiments that Chamberlin often inflicted on his students at Harvard University. A well-known leader of this group is Vernon Smith .
The motivation for experimental economics is to overcome the obvious fact that most mainstream neoclassical models are self-professed abstractions which employ simplifying assumptions whose realism is always open to question. Given that any typical economic explanation is of the form ‘if the world is of form X and people behave according to proposition Y, then we will observe phenomenon Z‘, the obvious questions facing any economist who claims to offer a positive explanation of economic phenomenon Z are: Is the world of form X? Do people in fact behave according to proposition Y? And do we observe phenomenon Z?
Since it is usually difficult to determine whether people actually behave according to proposition Y, almost all empirical research is concerned with world X and phenomenon Z. The usual approach is to build a model of the economy based on proposition Y and try to determine whether or not the model can be confirmed when confronted by the data available after the event. Unfortunately, the available data are seldom decisive in any direct way. Instead, many additional assumptions must be made and thus any conclusions reached are always conditional.
Harvard positivism offers a different approach. Rather than accept the limitation of available data (which are usually aggregative and thus open to many methodological questions), experimental economics proposes to create a real world situation in which the assumptions of the typical mainstream model are true with respect to the claimed form of world X. Specifically, the experimental economists attempt to construct a world which is in fact of form X and then determine whether the behavior implied by proposition Y is logically consistent with the experimentally observed phenomenon Z. The extent of the laboratory skill of the experimenter is always the sole determinant of whether the experiment represents a successful exercise in economic positivism.
The followers of Paul Samuelson’s methodology adopt a much less fundamentalist view of economic positivism. Samuelson is the famous Massachusetts Institute of Technology (MIT) economist who published his views about methodology in his PhD thesis [1947/65]. Those following his methodology, yet seeking to assure the optimistic promises of positivism, argue that the minimum condition for a positive contribution to economic understanding is that anyone’s positive theory must be capable of yielding to refutations based on positive evidence. In short, all truly positive theories are empirically refutable in principle. All that can be assured by such a weak requirement is that the proposed positive theory is not a tautology – as Terrence Hutchison  recognized in the late 1930s. It should be clear that this minimalist version of positivism is serving more the interests of mathematical model builders, who wish to avoid all of the menial unpleasantness of dealing with complex real-world empirical data, than the interests of those who are concerned with promoting truly positive economics. For many mathematical economists, the elegance of one’s model is always much more important than whether the model’s assumptions are empirically realistic or whether the model’s implications are useful with respect to economic policy.
Usefulness is the keystone of the positivism promoted by the followers of so-called Chicago school economics. However, there are two aspects of usefulness. On the one hand, providing positive theories that can be used as instruments by policy makers is one concern. On the other hand, being useful for promoting mainstream economics in general, and confirming beliefs in the omnipotence of the market system in particular, is another concern of the Chicago school.
In his 1953 essay, Friedman gives a compelling argument for why anyone who is only interested in providing useful theories for policy makers ought to eschew the typical philosophical prejudices associated with the group of analytical philosophers often called ‘logical positivists’ and instead recognize that questions concerning the verifiability, falsifiability, or even a priori realism of the behavioral assumptions of economic models are of much less concern than the usefulness of their results. As I explained in my 1979 Journal of Economic Literature article, it is easy to see that such an argument is really one favoring an instrumentalist methodology. The interesting question is, why would Friedman or anyone else see his argument as one promoting some form of positivism?
Friedman’s essay was not an argument against positivism but only one against the more sophisticated logical positivism. Positive evidence still matters for Friedman. His only restriction is to limit the evidence to that of results or predictions and thereby exclude a priori or logical analysis of models, assumptions and theories as a determinant of the usefulness of positive theories. Positive data obviously play an essential role in Friedman’s methodology. But for Friedman the only relevant positive data will be successful predictions which assure the usefulness of one’s model or theory. From a standpoint of methodology, there is nothing inherent in Friedman’s methodological essay that would prevent his form of instrumentalism from being used by Post-Keynesians or even Marxists.
When it comes to ideological questions, however, other members of the Chicago school are much more prominent. In 1977, George Stigler and Gary Becker offered a manifesto for those who believe in mainstream economics. Their argument, simply stated, was that they as Chicago school economists will offer models of the economy (i.e. of world X) which do not engage in analysis of the psychological (subjective) makeup of individual decision makers but instead offer analyses of the objective (positive) cost situations facing the individual decision makers and thereby explain any observable, positive behavioral evidence in question (i.e. phenomenon Z) – all observed changes in behavior will be explained as consequences of observable and objective cost situations.
Each positive economic model which succeeds (they never seem to report any failures) is offered as yet more confirming evidence that one can explain any social or behavioral phenomenon with an appropriately constructed mainstream model (i.e. where proposition Y incorporates assumed maximization behavior in a free-market system). For this branch of the Chicago school, the real purpose of mainstream neoclassical model building is once again to confirm the truth of a market-based system of social coordination [Boland 2003].
Stigler and Becker may be correct in promoting neoclassical economics as the only true explanation of social and individual behavior but, if so, it ought to be tested in a more critical manner. At the end of the 1950s, a group of London School of Economics (LSE) economists proposed a more critical approach to economic model building. While it is easy to find positive evidence to confirm anyone’s favorite model, the ‘scientific’ issue is one of approaching the evidence in a less predisposed manner. Such an approach does not preclude a priori beliefs; it merely cautions one to let the positive evidence do the talking.
The LSE approach to positivism was the self-conscious product of a group of young economists led by Richard Lipsey who formed what was called the ‘LSE Staff Seminar in Methodology, Measurement and Testing’. The seminar was to some extent inspired by Popper’s presence at LSE and his emphasis on criticism and empirical testing as the true basis for science. The message of the seminar was captured in Lipsey’s well-known 1960s textbook, Introduction to Positive Economics. The main thrust for Lipsey was the advocacy of developing an appreciation for real-world empirical data. His textbook became the major platform for all of modern economic positivism.
The combination of testing and measurement is the hallmark of LSE positivism. It is thus not surprising to find that econometrics plays a prominent role. But, unlike the instrumentalist tendency found among American econometric model builders, LSE econometrics is supposed to be helping us to assess any economic proposition that might arise. The positive/normative distinction was to play a central role since it was thought that all normative statements are untestable and thus ‘unscientific’.
Modern Economic Positivism is Profoundly Confused
As we have learned from historians of science (such as Thomas S. Kuhn and Joseph Agassi), most disciplines can be defined by their leading textbooks. The foundation of modern economic positivism continues to be Lipsey’s textbook, Introduction to Positive Economics. The evolution of this book closely reflects how the practice of positivism has developed over the last fifty years. However, if one examined the introductory ‘scope and method’ part of the first edition of Lipsey’s famous textbook, it would be difficult to understand how this book has become the foundation for modern economic positivism. Lipsey proudly announces that his book is about ‘POSITIVE ECONOMIC SCIENCE‘. The North American editions of his book play down the emphasis on ‘science’ (presumably because in North America such emphasis is considered pretentious) but then continue to share his emphasis on ‘positive’. Yet a careful examination of his 1963 book shows that empirical evidence can be decisive only in a negative way. Specifically, Lipsey parrots the part of Popper’s philosophy of science that claims that truly scientific theories can be refuted by empirical evidence but can never be verified by empirical evidence. In effect, then, according to Lipsey circa 1963, his book is really about NEGATIVE economic science!
This apparent inconsistency is abruptly corrected in his second edition, where he says he has
abandoned the Popperian notion of refutation and [has] … gone over to a statistical view of testing that accepts that neither refutation nor confirmation can ever be final, and that all we can hope to do is discover on the basis of finite amounts of imperfect knowledge what is the balance of probabilities between competing hypotheses. [Lipsey 1966, p. xx]
While this may accord better with common notions of science, it is not clear that there is anything positive (or negative!) left in the LSE version of positivism.
By the time of the sixth edition we are being told that only positive statements are testable. Normative statements are not testable because they depend on value judgements. Moreover, ‘statements that could conceivably be refuted by evidence if they are wrong are a subclass of positive statements’ [1983, p. 6]. So practitioners of positive economics ‘are concerned with developing propositions that fall into the positive, testable class’ [p. 7]. But, looking closer, on page 5 it is asserted that a statement is called ‘testable’ if it can ‘be proved wrong by empirical evidence’ and then turning to page 13 we are told it is ‘impossible to refute any theory conclusively’! Unless Lipsey meant something different from what appears on page 5, it would seem that the class of positive economic statements is empty and thus positive economics is impossible. If there is any doubt about whether the advocates of LSE positivism are profoundly confused about methodology, the 1988 Canadian edition of Lipsey’s book provides the proof: we are boldly told, ‘There is no absolute certainty in any knowledge’ [Lipsey, Purvis and Steiner 1988, p. 24]. I ask, how can one claim to know with absolute certainty that one cannot know with absolute certainty?
Their bold statement is self-contradictory and yet it appears to be the foundation of modern economic positivism. As is well known, anything can be proven with a foundation containing contradictions, and whenever it is possible to prove contradictory things the proofs are meaningless. Thus, we would have to conclude that nothing can be accomplished with the modern positivist’s methodology if that methodology is the one described in the various versions of Lipsey’s famous book. I think Lipsey should not have simply dropped Popper in order to avoid some ‘problems that seem intractable to a believer in single-observation refutations’ [1966, p. xx]. While his move will please those philosophers of science who are all too eager to dismiss Popper’s challenges to logical positivism, I think that Lipsey should have tried to critically examine those ‘intractable’ problems.
Positive Science or Positive Engineering?
Even though the philosophy of economic positivism has not been well thought out by its main proponents, it still captures all the satisfying notions that most mainstream economists seem to desire. On the one hand, it appears to support the commonly accepted view of explanatory science. On the other hand, it appears to support the appropriate cautions for a socially acceptable practice of social engineering. Specifically, both perspectives are served by the common view that positivism represents the avoidance of value judgments.
Positive Evidence About Positive Economics
Having discussed the nature of the economic positivism explicitly discussed in positivist textbooks, the next consideration ought to be about how positivism is actually practiced in positive economic analysis. The salient feature of all examples of ‘positive’ economic analysis is their conformity to just one standard format. Specifically, after the introductory section of a typical positive economics article there is a section titled ‘The model’ or some variation of this. This is followed by a section titled ‘Empirical results’ or something similar, and a final section summarizing the ‘Conclusions’. The question that should be considered is: why do virtually all positivist papers conform to this one format? Is the dominance of this uniformity the only success of modern economic positivism?
Explaining the Use of the Standard Article Format
Despite the ease one might have for confirming such a positive model of economic positivism (e.g. [Boland 2003, Chapter 12]), there is apparently no discussion of why papers should be written according to the observed format – apart from the recent discussion limited to the rhetoric of economic positivism [see McCloskey 1989]. Of course, there is no need to discuss the standard format if everyone agrees that it presents no problem and it is doing its required job. My general theory is that the reason why the format is not discussed is that its purpose is simply taken for granted. Taking things for granted is a major source of methodological problems and inconsistencies in economics, although the problems are not always appreciated. This is the case with the widespread use of one common format for mainstream empirical research papers. Perhaps there is no discussion because the job performed is merely one of an elementary filter, one which presumes that only papers that can be expressed in the standard format could ever make a positive contribution to positive economics. This presumption is also not discussed anywhere. So, just what is the purpose of the standard format?
While there need not be anything inherent in positivism that would connect its practice with the development of mainstream neoclassical economics, the two are closely related. The purpose of the standard format for those articles which purport to provide positive neoclassical economic analysis is exactly the purpose of promoting positivism in the first place. The purpose is the facilitation of a long-run inductive verification of knowledge even though the format is promoted by people who would see themselves practicing a more modest view of knowledge and method, a view which supposedly denies induction. At the root of this view is the conviction of Manifest Truth [see Boland 1997, Chapter 8]. More specifically, it is the conviction of mainstream economists that neoclassical economics represents a true theory of society – that the real world is manifestly what neoclassical economists claim it is – and thus any model based only on facts generated in the real world will in the long run lead one to see the Manifest Truth which in this case is believed to be the veracity of neoclassical economics. Basing models only on facts generated in the real world is, of course, the claimed purpose of positivism.
To understand the relationship between the standard format and the research program to verify neoclassical theory, we need to consider the following questions. What constitutes a successful positive analysis? What would be a failure? And, in order to determine what constitutes a success, it would seem that we ought to consider a more fundamental question: what is the objective of mainstream neoclassical model building?
If the usual published positive economics articles are actually considered contributions to ‘scientific knowledge’, then it can only be the case that the hidden objective of such positive economics is the one of Chicago positivism, namely, a long-term verification of neoclassical economics. Specifically, each paper which offers a confirmation of the applicability of neoclassical economics to ‘real world’ problems must be viewed as one more positive contribution towards an ultimate inductive proof of the truth of neoclassical theory. My reason for concluding this is merely that logically all that can be accomplished by the typical application of neoclassical theory to ‘real world’ phenomena is a proof that it is possible to fit at least one neoclassical model to the available data. Critics can always say that a model’s fit may be successful in the reported case but it does not prove that it will be successful in every case. I would argue that the agenda of positive mainstream neoclassical research programs presumes that if we can continue to contribute more confirming examples of the applicability of neoclassical economics, then eventually we will prove that it is the only true theory of the economy.
Positive Success or Positive Failure?
Clearly an examination of the format of a typical positivist economic analysis reveals that, as a form of rhetoric, economic positivism has been very successful. But has it been successful at fulfilling the broader promises of positivism? This is a particularly important question for those of us who reject the possibility of an inductive proof for any theory such as neoclassical economics.
While many of the proponents of the market system of prices in general and of privatization in particular are also proponents of positive economic analysis to support their views, it is seldom recognized that advocacy of either view is inconsistent with a non-romantic practice of positivism. It is not difficult to imagine the positivist economist’s response to a simple observation that, while the positivist economists base their analysis of economic phenomena on the presumed existence of a perfectly functioning market system of prices, the world outside our windows is not such a perfectly functioning system.
For example, if the world were governed by a market system of prices without governmental interference or private collusion, then eventually society’s resources would be optimally allocated according to the desires of all individual consumers. And, we are told that the world outside our window is in a state of equilibrium and, specifically, all prices are equilibrium prices. For this reason, any subsequent introduction of governments into the model will usually be seen to result in sub-optimal allocations of resources. Thus it is argued that privatization and the reliance on prices (as the only information appropriate for social coordination) is to be recommended.
To be fair, it should be recognized that the advocacy of privatization is a relatively recent phenomenon and not all advocates consider themselves to be positivists. Moreover, not all positivists advocate privatization despite what may seem to be the case today. During the 1950s and 1960s, most of the positivists were engaged in the advocacy of government interference in everyday economic affairs on the basis of what they called Keynesian economics. To these positivists it was enough to look outside our windows and see that the world is characterized by cyclical high unemployment and various levels of instability. Much of the academic effort in that period resulted in the development of the econometric approach to economic positivism which was intended to assist governments in the process of managing and ‘fine-tuning’ the economy.
It would seem that truly positive economists would shun such advocacy and simply and dispassionately explain the world the way it is. Namely, they should explain how phenomena are generated in a world where governments and collusion are commonplace. The obvious fact that many proponents of economic positivism are almost always engaged in the advocacy of simplistic engineering views such as either global privatization or governmental macroeconomic management should lead one to recognize that too often today economic positivism is mostly, and perhaps only, rhetoric.
Title image: Paul Samuelson by Ara Guler/ more
- Boland, L.  A critique of Friedman’s critics, Journal of Economic Literature, 17, 503-22
- Boland, L.  The Foundations of Economic Method (London: Geo. Allen & Unwin)
- Boland, L.  Critical Economic Methodology: A Personal Odyssey (London: Routledge)
- Boland, L.  The Foundations of Economic Method: A Popperian Perspective (London: Routledge)
- Comte, A. [1855/1974] Positive Philosophy, (New York: AMS Press)
- de Marchi, N. [1988a] Popper and the LSE economists, in de Marchi, N. (ed.) The Popperian Legacy in Economics: Papers Presented at a Symposium in Amsterdam, December 1985 (Cambridge: Cambridge University Press), 139-66
- Drury, H. [1922/68] Scientific Management, 3rd edn (New York: AMS Press)
- Friedman, M.  The methodology of positive economics, in Essays in Positive Economics (Chicago: University of Chicago Press), 3-43
- Harbury, C.  Descriptive Economics, 6th edn (London: Pitman)
- Hayek, F.  The uses of knowledge in society, American Economic Review, 35, 519-30
- Hutchison, T.  The Significance and Basic Postulates of Economic Theory (London: Macmillan)
- Jackson, R.  Rational Economics (New York: Philosophical Library)
- Keynes, J.N.  The Scope and Method of Political Economy, 4th edn (London: Macmillan)
- Lipsey, R.  An Introduction to Positive Economics, 1st edn (London: Weidenfeld & Nicolson)
- Lipsey, R.  An Introduction to Positive Economics, 2nd edn (London: Weidenfeld & Nicolson)
- Lipsey, R.  An Introduction to Positive Economics, 6th edn (London: Weidenfeld & Nicolson)
- Lipsey, R., D. Purvis and P. Steiner  Economics, 6th edn (New York: Harper & Row)
- Lundin, R.  Objective Psychology of Music, 2nd edn (New York: Wiley)
- McCloskey, D.  Why I am no longer a positivist, Review of Social Economy, 47, 225-38
- Samuelson, P.A. [1947/65] Foundations of Economic Analysis (New York: Atheneum)
- Smith, V.L.  Microeconomic systems as an experimental science, American Economic Review, 72, 923-55
- Spencer, H.  System of Synthetic Philosophy (New York: D. Appleton)