Under the unassuming heading "Other Considerations Concerning Simple Ideas," Locke next introduces one of the most important topics in the entire Essay: the distinction between primary and secondary qualities. Locke tells us that there is a crucial difference between two kinds of simple ideas we receive from sensation. Some of the ideas we receive resemble their causes out in the world, while others do not. The ideas which resemble their causes are the ideas of primary qualities: texture, number, size, shape, motion. The ideas which do not resemble their causes are the ideas of secondary qualities: color, sound, taste, and odor.
The best way to understand the distinction between primary and secondary qualities is in terms of explanation. Whenever you have the sensation of a square book the cause of that sensation is some sort of shape out in the world (though not necessarily squareness, since there may be some optical illusion, because distance, for instance, forcing you to perceive the shape incorrectly), so the explanation for sensation of shape is shape in the external world. Whenever you have a sensation of blue, on the other hand, the cause is not blueness out in the world. The cause is some specific arrangement of the insensible parts of matter. Explanations for secondary qualities refer only to primary qualities.
Locke's argument for this claim is based on his estimation of the "best science available", which he believes is Boyle's Corpuscular Hypothesis. According to the best scientific picture we have of the natural world, Locke argues, all that is out there are colorless, tasteless, soundless, odorless corpuscles of matter. Using only these indivisible bits of matter and their motions, we can explain not only our sensations of primary qualities, but our sensations of secondary qualities as well. Sensations of color, odor, taste, and sound are caused by the primary qualities of arrangements of matter. (Locke refers to these arrangements as the "powers" of objects to cause sensations.) Given that we are able to explain everything we need to explain by positing the existence only of primary qualities, he reasons, we have no reason to think that secondary qualities have any real basis in the world. An argument of this form is often called an "argument from parsimony" and rests on the premise that it is best not to posit the existence of explanatorily superfluous entities.
The rest of Locke's discussion of the primary/secondary quality distinction focuses on making the conclusion seem more plausible. He presents a number of thought experiments designed to bring our intuitions into line with his. First, he describes breaking a piece of wheat up into smaller and smaller pieces. He points out that as small as the wheat becomes we cannot conceive of it without its primary qualities (presumably since the very idea of a body without shape or size is incoherent) whereas we can conceive of the wheat without color (presumably because there is nothing literally incoherent about a body without color, even if it is difficult to imagine one in actuality).
He next considers an almond that is being pounded with a pestle. As it gets broken up into smaller and smaller pieces, the color changes from a pure white to a dirtier hue, and the taste goes from sweet to oily. Yet all that was altered was the texture of the nut. Clearly, he concludes, the secondary qualities depend on the primary qualities.
Finally, he takes the example of a flame. If we put our hand in the flame we have a sensation of pain. If we look at the flame we have a sensation of color. No one would claim that pain is in the flame itself, he points out, so why do we suppose that the color is?
Locke's argument for the claim that secondary qualities do not exist out in the world as we perceive them is obviously only as strong as the science it rests on. He knew that if that science turned out to be false, his entire argument would crumble. One might wonder why he would be willing to risk such an important point on an uncertain theory.
At least part of the answer to this question lies in the fact that the entire reason that the distinction between primary and secondary qualities was important to Locke was because of his belief in the *new mechanistic science*. Locke was intent on giving a philosophical clarification of the distinction specifically because he wanted to clear the way for the new science to take hold. He suspected that a general reluctance to believe in a colorless, odorless, tasteless world would prove to be a major obstacle to the acceptance of theories like Boyle's. He, therefore, wanted to make this stark view of the world more palatable, at least on a purely intellectual level (on a visceral, emotional level it remains hard to stomach even for modern physics students). If the science turned out to be wrong enough to leave his argument unsupported, then he would not have had much interest left in supporting his argument anyway.
Turning from the argument to the theory itself, it is difficult to conceive of how the motion of colorless, odorless particles is supposed to cause sensations like blueness and sweetness in us. Given that even today, with science as advanced as it is, we still cannot solve this mystery (it remains one of the most hotly debated problems in both philosophy of mind and in the cognitive sciences), it might seem unfair to hold Locke accountable for leaving us without a satisfactory answer. To his credit, Locke himself recognized the explanatory gap and tried to smooth over it by offering yet another thought experiment. This time he asks us to consider a knife. When a knife cuts flesh, it causes pain. We cannot imagine what it is about steel that leads to the sensation of pain, but yet no one doubts that it is the steel, and not any pain inside the knife, that causes the pain in us. Though this thought experiment does not clear up the mystery at all, it shows that even Locke realized the most puzzling consequence of the theory he espoused. In fact, he recognized it so fully that it plays a large role in the theory of knowledge presented in Book IV.
Another puzzling aspect of Locke's theory of primary/secondary qualities is the ontological status of secondary qualities. In what sense are these qualities supposed to exist independent of observers? If there were no observers around, would they continue to exist as powers in the objects, or would they simply cease to be? There seems to be a certain ambiguity in the way Locke uses the word "powers." Either a power could be an intrinsic property of the object in the world, or else a power could be a relational property that exists between the object and the observer.
If Locke means a power to be an intrinsic property of objects, then secondary qualities do have an entirely mind-independent existence, even if they do not exist as we perceive them. It seems likelier, however, that Locke conceives of powers as relational properties. The ability of an object to cause certain sensations depends on the way in which the insensible constituents of matter interact with our own sense organs. For an object to have a power, then, should involve a certain relation between the object and the perceiver--powers should have as much to do with the laws of neurophysiology as with the laws of physics. If this is the case, then secondary qualities cannot be said to exist entirely independently of all observers. There needs to be at least the potential for their being observed (such as their being on a planet with creatures capable of perceiving them) in order for them to exist.
Locke's analogy between sweetness and pain provides further support for this reading. In the analogy, Locke claims that sweetness is in food exactly as pain or sickness is in food. Both are powers in the food to cause certain sensations in us. If sweetness really is in food in precisely the same way that pain is, then sweetness is certainly mind-dependent. No one would claim that pain still existed in a world devoid of creatures with pain receptors. Presumably, then, on Locke's picture, in a world devoid of creatures with color, sound, taste, and odor receptors, color, sound, taste, and odor would not exist.