Skip to Navigation Skip to Content Skip to Search Skip to Site Map

Hypotheses versus Queries in Newton’s Opticks

Kirsten Walsh writes…

A while ago I argued that the queries in Newton’s early optical papers are not hypotheses.  Rather, they are empirical questions that may be resolved by experiment.  In Newton’s Opticks, however, his queries become increasingly speculative – especially the famous ‘Query 31’.  What should we make of this?  Did Newton abandon his early distinction between hypotheses and queries?

In his early optical papers, Newton explains that “the proper Method for inquiring after the properties of things is to deduce them from Experiments”.  Having obtained a theory in this way, one should proceed as follows:

  1. specify queries that suggest experiments that will test the theory; and
  2. carry out those experiments.

He tells us that hypotheses have a role in this procedure.  They may be useful for: (a) suggesting further experiments, as the first step toward specifying queries; and (b) ‘illustrating’ the theory to assist understanding.

The queries in Newton’s Opticks have been much talked about, and often Newton has been accused of slipping hypotheses into his work under the guise of the more-respectable query.  To examine this claim, I looked at the draft manuscripts* of Newton’s Opticks; in particular, “The fourth book concerning the nature of light & ye power of bodies to refract & reflect it” (Add. 3970, 337-8).

The draft begins, as many of the other books of Opticks begin, with a list of observations, followed by numbered propositions.  However, it contains little in the way of argument and virtually no discussion of experimental evidence.  Shapiro points out that this is because this is a draft of an outline or plan of a book; not a draft of the book itself.  The propositions are things that Newton hoped to prove.  For example:

    Prop. 1.  The refracting power of bodies in vacuo is proportional to their specific gravities.
    Prop. 2.  The refracting power of two contiguous bodies is the difference of their refracting powers in vacuo.

The draft contains a section entitled ‘The conclusion’, which contains five ‘hypotheses’.  I am interested in ‘Hypothesis 2’:

    As all the great motions in the world depend upon a certain kind of force (wch in this earth we call gravity) whereby great bodies attract one another at great distances: so all the little motions in ye world depend upon certain kinds of forces whereby minute bodies attract or dispell one another at little distances.
    How the great bodies of ye earth Sun moon & Planets gravitate towards one another what are ye laws of & quantities of their gravitating forces at all distance from them & how all ye motions of those bodies are regulated by those their gravities I shewed in my Mathematical Principles of Philosophy to the satisfaction of my readers: And if Nature be most simple & fully consonant to her self she observes the same method in regulating the motions of smaller bodies wch she doth in regulating those of the greater… The truth of this Hypothesis I assert not because I cannot prove it.  But I think it very probable because a great part of the phaenomena of nature do easily flow from it wch seem otherways inexplicable…

I. Bernard Cohen describes this as “a ‘whale’ of an hypothesis” – and he’s right!  When Newton started writing out this statement, he intended for it to be ‘Proposition 18’.  But at some point, he has scratched out ‘Prop 18’, and re-branded it as ‘Hypoth 2’.  There is no real semantic difference between a proposition and a hypothesis, but, for Newton, there is an epistemic difference.  Propositions are things that he is able to assert as true.  Hypotheses are things that he is unable to assert, because he does not have the evidence.  Newton clearly hoped to assert Proposition 18.  But as he started to explicate it, he must have realised that he couldn’t prove it.  Thus, he re-labelled it as a hypothesis.

When Newton abandoned the fourth book, and restructured the rest of his Opticks, this ‘Hypothesis 2’ appears to have been re-worked to become ‘Query 31’ in Opticks, 2nd edition (1717):

    Have not the small Particles of Bodies certain Powers, Virtues, or Forces, by which they act at a distance, not only upon the Rays of Light for reflecting, refracting, and inflecting them, but also upon one another for producing a great Part of the Phaenomena of Nature?  For it’s well known, that Bodies act one upon another by the Attractions of Gravity, Magnetism, and Electricity; and these Instances shew the Tenor and Course of Nature, and make it not improbable but that there may be more attractive Powers than these.  For Nature is very consonant and conformable to her self…

Here, there is an obvious semantic shift between hypothesis and query: the query is stated as a question.  Some scholars have argued that this is the only difference between hypotheses and queries: in the Opticks, queries are simply Newton’s way of getting around his self-imposed ban on hypotheses.  I claim that there is more to the shift than this.  Newton is using the semantic structure of the query to explore a possible future research program.  The epistemic difference between the query and the hypothesis is similar to the epistemic difference between Popper’s falsifiable and unfalsifiable theories.  The former is testable-in-principle, whereas the latter is not; and testability is a necessary condition of something becoming well-tested.

There is a difference between Newton’s early queries and his later queries: the former are part of the process of justification; but the latter are part of the process of discovery.  In a previous post I noted that:

    While Newton’s [early] method of queries is experimental, it does not appear to be strictly Baconian.  For the Baconian-experimental philosopher, queries serve “to provoke and stimulate further inquiry”.  Thus, for the Baconian-experimental philosopher, queries are part of the process of discovery.  However, for Newton, queries serve to test the theory and to answer criticisms.  Thus, they are part of the process of justification.

The queries in Newton’s later work seem closer to the Baconian tradition that inspired him.

That the themes of Hypothesis 2 and Query 31 appear in Rule 3 of Principia, raises questions about the status of Newton’s ‘Rules of Philosophising’ and how we should interpret the re-branding of ‘hypotheses’ as ‘rules’ in later editions of Principia.  I’d love to hear what you think!


* Recently, Cambridge University put Newton’s papers online, making it possible for those of us who live ‘down under’ to examine copies of many of Newton’s manuscripts!

7 thoughts on “Hypotheses versus Queries in Newton’s Opticks

  1. I didn’t understand the answer in the crucial paragraph. Aren’t hypotheses intended to explore a possible future research programs also? How is the analogy between query/hypothesis, testable/non-testable supposed to run? Are you thinking of the hypotheses that Newton says he isn’t framing as non-testable hypotheses?

  2. Great post Kirsten! What an exciting find seeing the crossed out original.

    Given Newton says in the initial hypothesis “not because I cannot prove it”… does this indicate that a hypothesis could exist in similar terms to a theory, but simply be semantically different for the hypothesis purposes you identify?

  3. Hi Mike,

    Thanks for your questions!

    1) When Newton wrote his ‘hypothesis explaining the properties of light’ in 1675, he was using hypotheses to explore possible future research programs. But it seems that by the 1700s (at least, c. 1706-1717), Newton was no longer using hypotheses in this way. Instead, in Principia, ‘hypotheses’ are plausible, but unfounded, assumptions or conjectures. They seem to function as ‘helping premises’ in conjunction with propositions, so that their consequences and implications may be examined.

    2) My analogy between queries/hypotheses and testable/untestable is a new idea, and I’m not sure how far it will run. (Prima facia, the analogy doesn’t work at all, since questions have no truth value, and so are untestable. But hopefully we can ignore that for now!) For Newton, the truth or falsity of a hypothesis is irrelevant, since he thinks that hypotheses are underdetermined by the evidence. But queries are intended to explore speculative conjectures in order to see what their experimental consequences might be. These experimental consequences are, ideally, testable. We know that we shouldn’t believe a theory just because it is testable, but we know we should grant it higher epistemic status than a theory that is logically untestable. Similarly queries describe experimental consequences that are testable, but untested. So, while we shouldn’t believe any of the conjectures it explores (until they are ‘proved’ by rigorous experiment), it is still on higher epistemic footing than the hypothesis, for which there are no possible experimental consequences.

    Does this clarify things a bit?


  4. Hi Simon,

    Thanks for your comment. Hypotheses and theories are semantically similar, since they are both stated as propositions (as opposed to questions, commands, etc). For Newton, the difference between hypotheses and theories is an epistemic one: theories are certain; hypotheses are not. For a more detailed response, have a look at my earlier post on Newton’s corpuscular hypothesis.


  5. Thanks for the reply, Kirsten. How does your second answer (Newtonian hypotheses aren’t testable) fit with your first answer (Newtonian hypotheses have consequences when conjoined with propositions)? Also, how does your account of Newtonian hypotheses as non-testable hypotheses fit with what seems to be his definition (quasi-definition?) in the the general scholium? “Quicquid enim ex phænomenis non deducitur, hypothesis vocanda est” “For whatever is not deduc’d from the phænomena, is to be called an hypothesis” as Motte has it.

  6. Hi Mike,

    1) I don’t think there’s any obvious inconsistency or lack-of-fit between my two answers. I don’t think Newton thinks his hypotheses are necessarily logically unfalsifiable. Rather, they are at least (what I would call) ‘methodologically unfalsifiable’. That is, Newton is not interested or concerned with testing these hypotheses. But it wouldn’t be inconsistent to argue that they are indeed logically unfalsifiable, since the idea that hypotheses only have observable consequences when they are conjoined with other assumptions/propositions, and hence, no individual hypothesis is ever falsifiable, is a simple (and uncontroversial?) Duhemian point.

    2) In the General Scholium, Newton doesn’t give a full definition (i.e. necessary and sufficient conditions) for ‘hypothesis’. He simply gives a sufficient condition for ‘hypothesis’ – this is because he is attacking a particular kind of hypothesis. From the various (published and unpublished) methodological statements Newton makes over time, it appears that he has a disjunctive definition of ‘hypothesis’:
    H1. Something that is, at best, only approximately true;
    H2. A conjecture or speculation – something not based on empirical evidence; or
    H3. A causal explanation – something concerning the nature of the phenomenon, rather than its physical properties.
    I think both my account of hypothesis and Newton’s definition in the General Scholium fit this disjunctive definition.

    Thanks for helping me to ‘nut out’ these claims!


  7. Sure. You’re doing important work here.

    Make sure that your claim that Newtonian hypotheses aren’t falsifiable doesn’t become trivial by linking it too closely with the Duhem-Quine thesis.

    Also: if some propositions that aren’t deduced from the phenomena are testable, and Newton gives a sufficient condition for hypotheses in the general scholium, then some hypotheses will be testable.