Democritus, answered from the texts and cited to the page.
Sweet, bitter, hot, cold, color, sound, smell — these do not exist as objective properties of things. They are what atomic interactions produce in our sensory apparatus when the eidōla (films) streaming from objects strike sense organs whose atomic structure determines the response. The things themselves have only the primary properties of their constituent atoms: size, shape, motion, arrangement.
The phenomenal qualities are nomoi — conventional names we give to atomic effects on us. The reality is atoms and void. Fragment D16 puts the doctrine in its most condensed form: "by convention sweet, by convention bitter, by convention hot, by convention cold, by convention colour; but in reality atoms and void." The Democritean coinage eteē (in reality, in truth — Galen attests this is my word) marks the distinction between the conventional-phenomenal realm and the real-atomic realm.12
The eliminativism is strict, not relational, not two-aspect: secondary qualities have no reality in the things; they exist only in the affection of perceivers by atomic structures.3 The softer relational and two-aspect readings preserve a phenomenal-realism that the fragments explicitly deny.45
We thus have a dichotomy between how things seem to us and how they are in reality, expressed in the celebrated slogan (D16) 'By convention sweet... in reality atoms and void.' Further, the distinction between the reality of things and the appearances which that reality presents has to be supplemented by an account of the 217 Doctrines: Epistemology causal processes via which we receive those appearances.The Atomists, Leucippus and Democritus, p. 204
The overthrow is a fall for you' [= D23]. d. GALEN On the Elements according to Hippocrates 1.2 'For by convention colour, by convention sweet, by convention bitter, but in reality atoms and the void' says Democritus, who thinks that all 131 For the translation see note on D20. 132 A follower of Democritus; see DK 76.The Atomists, Leucippus and Democritus, p. 131
Was that thought, then, totally inconsistent? Or can the appearance of systematic contradiction be eliminated or at least mitigated? The former interpretation is based on the atomists' account of the secondary qualities (see above).The Atomists, Leucippus and Democritus, pp. 203–204
And in that case, what ground could there be for denying that they actually do have some colour? The reasons for denying the ascription to the atoms go deeper than the contingent fact that all the atoms which human beings encounter are too small to be perceived by human beings.The Atomists, Leucippus and Democritus, pp. 163–164
On that conception of an intrinsic property, in which an intrinsic property is an essential property of a kind of thing, it remains to be answered why some secondary quality (or qualities) should not be an intrinsic property (or properties) of an atom.The Atomists, Leucippus and Democritus, pp. 164–165
The interpretation of the nomos/physis distinction is the central question in Democritean epistemology and one of the foundational questions in the longer history of primary/secondary qualities. The strict eliminativist reading is most directly supported by the verbatim D16 and is the standard modern consensus (Taylor 1999, Furley 1987, Barnes 1979, KRS 1983). The relational-realist reading has been defended in various forms by readers who emphasize that even on the strict reading the phenomenal qualities exist as the affection of perceivers — i.e., relationally. The two-aspect reading emphasizes that Democritus preserves the phenomenal realm as a real domain even while denying it independent existence. The relativity-of- perception fragments (D17, the dependence on bodily condition) have been used to support both Pyrrhonist appropriations (Sextus) and modern relational readings. Locke explicitly credited atomism with the primary/secondary distinction that shaped modern philosophy. The 'wretched mind' fragment (D23) raises a related but distinct question (the epistemological tension between mind and senses) that bears on but does not settle the eliminativist question.