Practical analysis means cutting a thing to pieces or dissolving it, and this has a certain value because it multiplies objects. But it does not increase our knowledge of the first thing. On the contrary, by destroying a thing we render a knowledge of it impossible. The analysis which gives knowledge is Metaphysical Abstraction--an attention concentrated on the parts of a thing without destroying their connection with the other inherent parts. The metaphysical elements may be quite different from the mechanically divisible parts. They are generally a species of things which could not exist alone, such as red, blue, straight, curved, square, round, acid, sweet, insipid, fragrant, sharp, hot, heavy, dull, loud, bright, and a multitude of properties of that abstract kind. For many of these--at least for the description of them--a comparison of two or more things is essential. A sound is heard to be loud by comparison with another which is low or soft; a knife is known to be blunt by experience of another more sharp, or the same knife in a sharper condition. But comparison does not alter the essential character of abstract attention--it serves merely as an incitement to it. Difference between qualities otherwise alike whets our attention to a finer discrimination. The properties recognised by each sense are easily distinguished in the bulk from those of another sense. Colour is distinct from Figure in a more marked degree than red from blue or square from circular. Fine degrees of Sound may be difficult to discriminate, but not the difference between a sound and a smell or a taste. Still broader contrasts give rise to an artificial but sometimes useful kind of attribution--the negative. When we do not know much concerning the positive characteristics of a thing, it is something to know that it has _not_ this or that property. What Thought is, positively, few people know, but they are able to say (with a little prompting) that it is un-extended, im-material, im-ponderable, and so forth. This comparison re-acts on the thing better known, and so we call visual objects 'extended' from their dissimilarity to thoughts. But for that there would have been no occasion to notice the abstract extension of visual objects. The term 'visual object' would have tacitly included extension. There must be great and general ignorance of a thing to excuse the negative attribution: it is not allowable to speak of plants as non-metals, or sheep as non-horses, but a large class of animals is called in-vertebrate. In this case the negative property serves to bar a possible inference that all animals are vertebrate, since those we know best are so.