Concerning inference, of which I shall have more to say in the next chapter, I have now to note that it is of two kinds: first, perceptual inference, or inference from direct experience; secondly, conceptual inference, or inference based on experience, but reached through the exercise of the reasoning faculties. The latter involves the process of analysis or isolation; the former does not. There is a marked difference between the two. Perceptual inferences are the outcome of practical experience, but do not go beyond such practical experience. Conceptual inferences are also based on experience, but they predict occurrences never before experienced. Perceptual inferences, again, deal with matters practically; but conceptual thought explains them. The expectation of a storm when the thunder-clouds are heavy is a case of perceptual inference. It is the outcome of a long-established association, and is not reached by a process of reasoning involving an analysis of the phenomena. But if, though the sky is clear, a west wind and a rapidly falling barometer lead me to predict rain, the inference is conceptual, and gained by me or for me by a process of reasoning; for the barometer was the outcome of the analysis of phenomena. In the mind of the rough sailor-lad, however, the fall of the mercury and the succeeding storm may be connected by mere perceptual inference, the phenomena being simply associated together. If, however, there is any attempt at explanation, correct or incorrect, there is so far a conceptual element. In a little fishing-village on our south coast, a benevolent lady presented the fishermen with a Fitzroy barometer. I happened shortly after to remark to one of the men that the summer had been unusually stormy. "Yes, sir," he said, "it has. But then, you see, the weather hasn't no chance against that new glass." Here there was an attempted explanation of the phenomena. The falling glass was conceived as somehow causing bad weather. It is hard to draw the line between perceptual and conceptual inferences, or rather to say, in this or that case, to which class the inference belongs, because man, through language, lives in a conceptual atmosphere. Moreover, the same result may, in different cases, be reached by perceptual or by conceptual inference. A child who had seen a great number of ascending balloons might, on seeing a balloon, expect it to ascend by a perceptual inference; but a man, knowing that the balloon was full of a gas lighter than air, might expect it to ascend through the exercise of conceptual inference. And just as in adult civilized life our constructs have more and more conceptual elements built into them, so do our inferences become more and more reasoned. It is probable that in an adult Englishman every inference has a larger or smaller dose of the conceptual element.