This investigation will show that intonations, prosodemes and ‘secondary phonemes’, pitch and stress morphemes and phonemes, and suprasegmental features in general, can all be obtained as a result of the single operation of analyzing the utterances of a language into simultaneous components. It will show that the various limitations of phonemic distribution, including defective distribution of phonemes, can be compactly expressed by means of the same operation. When this operation is carried out for a whole language, it breaks all or most of the phonemes into new sub-elements (components). Each of the old phonemes will be a particular simultaneous combination of one or more of these new elements; and the total number of different components will be much smaller than the previous total number of different phonemes. It will be possible to select and symbolize the components in such a way as to show immediately the limitations of distribution, and in many cases the phonetic composition, of the phonemes in which they occur.
- Harris (1944), a pag.181

The traditional phonemes indicate explicit physical events: time-stretches of sound (sound-waves), or sets of simultaneous motions of the ‘vocal organs’. The new component elements also indicate explicit physical events: time-stretches of sound-waves, or motions of particular vocal organs. The only difference is that phonemes are elements which can, in general, occur only after one another, while components are elements which can also occur simultaneously with each other (as well as after each other).
- Harris (1944), a pag.186

[…] it will be seen that in special cases a component (or its phonetic value) may extend over more than one phoneme; and it will be important to note what happens when we get such LONG COMPONENTS […]. The work that a component can do in the description of a language depends on its length. Components whose length is that of one phoneme can be used to describe the phonetic composition of phonemes […] or the dissection of a single allophone into two or more phonemes […]. Components whose length is that of two or three phonemes (or thereabouts) can be used to indicate the limitations of distribution of any phoneme which contains them […]. And components which can extend over long sequences of phonemes are used in the descriptions of intonational and other contours […].
- Harris (1944), a pag.187

Since the components are to identify phonemes, or more generally speech sounds, each component must have a stated phonetic value in each environment in which it occurs. As in the case of phonemes, there is no reason to require that its phonetic value be identical in all the environments. The component can therefore have different phonetic variants (allophones) in various positions, and the environmental factor which determines the particular allophone may be anything outside the component itself: other components with which it is concurrent, neighboring components or pauses, position of the component within the sequence of segments, etc.
- Harris (1944), a pag.187

Again as in the case of phonemes, it is not required that components have a constant phonetic value throughout their duration. A component may have a phonetic value which changes in a fixed way in respect to its end-points: e.g. falling tone, increase in nasality, voiceless beginning and voiced ending.
- Harris (1944), a pag.187

[…] there is no reason for us to restrict every component to the length of one phoneme. If a component is always common to a sequence of phonemes, we can say that its length is the length of the sequence.
- Harris (1944), a pag.188

[…] the components are generalized phonemes: they appear concurrently with each other as well as next each other, and they may have a length of several phoneme-places as well as of one phoneme-place. And when we write with these components it is natural that various ones will have various lengths; each of them has to have some stated lengths, and the components symbolized by ¯ and ̿ are simply among those that in some situations have 2-phoneme length.
- Harris (1944), a pag.192

The components are essentially similar to phonemes in that both are distributional symbols with phonetic values. That is to say, the observed physical events are always sounds, and the criteria for classifying them into linguistic elements—whether phonemes or components—are always distributional.
- Harris (1944), a pag.203

The components are merely generalizations of the phonemes, extending the very development which gives us phonemes out of sounds. In writing allophones we have one distinguishable sound per symbol (hence closely abiding by the physical event); but there are many symbols and each usually has a highly restricted occurrence. In writing phonemes we often have several distinguishable sounds per symbol, usually but not always having considerable phonetic similarity (hence abiding rather less closely by the physical event); but there are fewer symbols with a wider distribution for each. In writing components we usually have more distinguishable sounds per symbol, sometimes with no common feature (hence abiding much less by the physical event); but there are fewer symbols yet, with much wider distribution for each.
- Harris (1944), a pag.203