Sigismund has asked whether imitation of singing may not serve as an introduction to language lessons. He says: “The first real imitation which I observed in my boy was not repetition of articulate speech, but of a musical tone. When he was fourteen months old and had as yet imitated nothing (?), I occasionally sang to him a popular song whose melody began with a downward quarter (F-C), which interval recurred frequently and forcibly in the song. I was greatly surprised when the child, though very drowsy, sang this measure correctly, an octave higher. The following day the same thing happened, and this time without any example.... Is it the rule or the exception that the infant sings imitatively before he speaks so? Many mothers whom I have questioned were uncertain whether such singing had occurred at all, but they had probably simply failed to notice it. The result of my own investigations and observation points to the probability that children, like birds, more easily comprehend and repeat singing tones than speech.”[554] Ufer justly replies to this that while children do indeed often sing before they can talk, we have no reason to affirm that this is the rule. The child observed by Miss Shinn, for example, first made feeble efforts to imitate singing in its fortieth month.[555] It is always unsafe to attach too much importance to isolated cases. It is characteristic of man that many of his inherited capacities are left afloat, as it were, and must be anchored by individual experience, thus affording opportunity for the development of varied individuality. Consequently, it is hardly possible to be too cautious in drawing conclusions for phylogenetic evolution from ontogenetic development.