would be used throughout, and the speaker would point to the parts of the diagram being referred to at the appropriate moments. The stumbling-block is the translation of an act of pointing into a diagram. Once we have drawn an arrow, we cannot erase it in a way which corresponds to the withdrawal of one's hand; we have to re-draw the diagram. And the arrows also clutter the diagram. because they are too like part of it. Different colours would help. Another use of words has been to suggest new classifications to the reader: for example, that a straight line may be considered as a particular kind of angle. This can also be shown visually. It takes more space, but is more vivid. There is a certain resemblance to a strip-cartoon, and if one has the resources and ability to translate this into computer graphics, the visual presentation can retain all its advantages. What would be the stages of such an animation? The following is one possibility. Note that the first figures represent the data. Here we use letters as a substitute for pointing. When the letters are found in the (verbal-algebraic) proof, we then have to find these letters in the diagram, and this tells us where to look. This is neater than the long arrows used on page 97 and saves re-drawing the diagram. Which is the easier to follow, the reader must judge for himself. this too could with great advantage be translated into computer graphics. How does the 'purely visual' approach cope with more complex proofs? Space must limit us to one further example—a proof of this more general theorem already referred to. Is this clearer than a verbal-algebraic proof (for which, see any traditional school geometry text), or is it another case of 'look, no hands'—-this time, no words? Since individuals differ in their preferences for visual or verbal-algebraic symbolism, there may be no general answer to this question. At present the latter system has achieved dominance. The chief purpose of the foregoing has been to question this fait accompli and examine the particular contribution of visual symbolism. Historically, one of the happiest marriages of the two systems is that due to Descartes (1569—1650). Any point in the plane of the paper is specified by its distances from two (usually perpendicular) lines. that is, by two numbers, written as an ordered pair. These coordinates, as they are called, may be positive or negative. A variable point corresponds to a pair of numerical variables; and a set of points with a given characteristic property, for example, that their distance from the origin is always equal to r, is_represented by an_equation satisfied by all the pairs of coordinates (x, y). By these means curves which are difficult to draw accurately can be represented algebraically: for example, an ellipse, which is the shape of a planet's orbit round the sun; a parabola, which is the shape a reflector must be to give a parallel beam (as for a car headlight) or to concentrate distant rays to a point (as for a radio telescope). Both general and metrical properties can be dealt with in this way: general properties, by using general relations between variable coordinates, and metrical properties, by giving particular numerical values to these variables. What this algebraical treatment of geometry adds is great power of manipulation and accuracy far beyond what is available by accurate drawing to scale and measurement of the drawing. But we still need the drawing to show what the set of points looks like as a whole. It is, for example, not obvious from the equations that the curve represented by y2=4a disappears into the distance in two directions, or that the curve represented by 7a+5a joins itself again, or that a simple change of sign in the latter will give us something looking completely different. That neither kind of representation is superior in all ways is suggested by the fact that we often use the method in reverse. Instead of starting with a known curve (all the above were known to Greek geometers about eighteen centuries before Descartes) and representing it algebraically, we may start with an algebraic concept, that of a function, and represent it graphically. The idea of a mathematical function is one of great generality. Broadly speaking, functions tell us how the objects in one set correspond to those in another: for example, how the distance travelled by an object may be found if we know the time; how the current through a given circuit may be determined if we the voltage. Functions may be represented in a variety of ways, including equations and graphs. For finding individual correspondences, an equation is very convenient. For example, if d metres is the distance travelled by a body in free fall under gravity (neglecting air resistance) and t seconds the time it has been falling, then d = 4•9 P. So the distance fallen after one second is 4•9 X 4 metres, after two seconds it is 4•9 X 4 metres, and so on. By taking (t, d) as Cartesian coordinates, we can show graphically the function as a whole.
THE TWO SYSTEMS COMPARED
Tentatively, we may now attempt a summary of the contrasting, and largely
complementary, properties of the two kinds of symbol. Visual Verbal-algebraic Abstracts spatial properties. Such as Abstracts properties which are independent Shape, position of spatial configuration, such as number
Harder to communicate Easier to communicate
May represent more individual thinking May represent more socialized thinking
The communicable, socialized properties of the verbal-algebraic system have doubtless
contributed to its predominance over the visual system. Yet whenever we want to represent also the overall structure of some topic, argument or situation, visual symbolism returns, as in organization charts (from firms to football teams), flow diagrams and family trees. The value of visual symbolism is also shown by the way in which it superimposes itself on the verbal - algebraic, in the form of spatial arrangement of written symbols. Auditory symbols are inevitably sequential in time. When written down, they are present simultaneously, the sequential arrangement being restored by scanning them in a conventionally agreed order. But this order may be departed from whenever we like. We may look quickly at the beginning and conclusion of an argument before examining details. We may recapitulate whenever we wish, and this becomes necessary more often as the argument becornes more involved. In other words, a verbal-algebraic exposition, once written down, shows the overall structure in addition to the logical-sequential implications within the structure, and it may be scanned in other ways besides the conventional left to right, top to bottom, order. Spatial symbolism finds its way into every detail of the verbal-algebraic system. 1. The position of a digit helps to show what number it represents. 2. Position shows which number is subtracted from which, 3. or divided by which. 4. Position shows correspondence between two sets, as in this proportion. 5. Its spatial arrangement is an essential property of a matrix. Many other examples could be given. Before concluding this chapter, it will be interesting to return briefly to the individual differences in imagery noticed by Galton and mentioned at the beginning. If we are right in thinking that visual imagery is that most favourable to the integration of ideas, and if it is not accidental that when we first become aware of how ideas relate to each other, we refer to the experience as insight, not as inhearing, then we might reasonably hypothesize that persons who have been noteworthy for their contributions to mathematical and scientific understanding will be found to use visual rather than auditory imagery. We may as well begin the list with Galton himself, who tells us that his own visual imagery was clear but that he lacked verbal fluency. Einstein (1879— 1955), in a letter to Hadamard, states that his preferred imagery is visual and motor, and that conventional words or other signs have to be sought for laboriously, only in a secondary stage' (Hadamard, 1945). A famous non-mathematical example is that of Kekule (1829—96), whose conception of the ring structure of the benzene molecule came to him in a dream where he saw a snake taking hold of its own tail. And the Nobel Prize winner Bragg (1890— 1971), in a television programme honouring his eightieth birthday, stated that his new ideas came to him for the first time in the form of visual images. This list is a partial and selective one, and we lack the comprehensive information about other famous mathematicians which would support or refute the hypothesis. An interesting discussion, along more general lines, of personality traits of mathematicians and others is to be found in Appendix 2 of Macfarlane Smith's Spatial Abiliy; where much other material of interest in the present context may also be found. Analysis, logical argument and socialized thinking are, rightly, much valued in mathematics, but we also need synthesis, insight and individual thinking. To some extent the former seem to be capable of being taught; the latter, at present, can only be sought. If we can discover more about the functions of the two kinds of symbol discussed in this chapter, and become one skilled in choosin an using them, this might well help us to develop and relate these two comlementary aspects of our mathematical thinking.
the theorem given last tells us that this angle is twice the size of thus angle