of all the members of the infinite class in example (3) above.
If we write
2 + 4 + 8 + 16 + 32 + 64 + 128 + 256 + ... ,
it is evident at once that as we go on adding successive terms, the sum rapidly becomes larger and larger. Actually, it is not
12 + 4 + 8 + 16 + 32 r'IG. 74. The sum 2 + 4 +8+ 16 + 32 + 64 +128 + 256·· ·increases without limit
enough to say simply "larger and larger." We must be more
precise. Let us note that by going out far enough in the series, we can make the sum of all the terms up to that point exceed any finite number, however large. This fact is indi- cated graphically in Figure 74. For example, if someone names the finite number 1000, we can, by taking 9 terms, make the sum 1022. If he raises the bid to 1,000,000, we can make the Paradoxes of the Infinite 121
sum 1,048,574 by taking 19 terms. If he cares to go as high as
1,000,000,000, we have only to take 29 terms to make the sum 1,073,741,822. No matter how large a finite number our imaginary adversary sees fit to choose, it is evident that we can always make our sum exceed his number by taking a suffi- ciently large finite number of terms. This is what the mathe- matician means when he says that "the sum of this infinite series is infinite." But now let us return to the problem of motion from one point to another. Suppose the distance from P to Q is 100 yards, and that we walk at the rate of 100 yards per minute. Then the time required for the first stage of the journey- half the distance from P to Q-is ~ minute; that for half the remaining distance, U minute; that for half the distance then remaining, Ys minute; that for half the distance then remain- ing,7l6 minute; and so on. In other words, the time in minutes required to go from P to Q is the sum of the infinite series 1111111 1 "2 + "4 + "8 + 16 + 32 + 64 + 128 + 256 + .... (Note that this is the sum of all the members of the infinite class in example (4) above.) Is the sum of this infinite series infinite? As in our previous series, the sum does get larger and larger as we go on adding successive terms. But it is not true that we can make the sum exceed any large finite num- ber which anyone cares to name. A glance at Figure 75 shows us intuitively that the sum approaches more and more nearly to 1, but never exceeds it. More precisely, if anyone names a finite number, however small, we can, by taking a sufficiently large number of terms, make the difference between our sum and 1 smaller than the named number. For example, if some- one chooses the number 10~0' we can, by taking 10 terms,
make the sum differ from 1 by 10~4· If he lowers the bid to
1,00~,000 ' we can make the sum differ from 1 by 1,04!,576
1 by taking 20 terms. If he cares to go as low as 1,000,000,000 '