Dry-climate tests.— In the work of 1907 (published 1909) upon the first group of 25 yellow pines from 1700 to 1900 A. D., several long sequences of variation in a 5 to 6 year period were noted. These were compared with rainfall records at Prescott and in southern California and the crests of rainfall and growth appeared to coincide in date. It was then seen that the temperature curve of southern California had a period and phase corresponding to the rainfall curve, but with the second minimum almost entirely suppressed, and that finally this temperature curve resembled in form and phase the inverted curve of sunspot numbers. In connection with the publication referred to (1909), a set of curves was prepared to show these relationships. This set is partly reproduced in figure 34, page 104. In the original drawing the tree-curve was the least satisfactory, which was to be expected, as no real certainty in the dating of rings existed at that time. After cross-identification the tree-curve was again integrated for the 11-year period and far better results were obtained. This new curve is given in the figure referred to.
This type of integrated curve gives many facts in a very condensed form. A differential or detailed form of presentation should accompany it, as in figure 25, showing the full series of individual observations and beside it the curve with which it is to be compared. The differential study of the Arizona trees will be taken up in connection with cycles, but can be summarized in the statement that in the last 160 years 10 of the 14 sunspot maxima and minima have been followed about four years later by pronounced maxima and minima in the tree-growth. Also, during some 250 years of the early growth of these trees, they show a strongly marked double-crested 11-year variation.
Wet-climate reaction. — In the very first group of continental trees studied, those obtained at Eberswalde near Berlin, the remarkable fact was recognized at once that 13 trees from one of those carefully tended German forests show the 11-year sunspot curve since 1830 with accuracy. The variation in the trees is shown in plate 8. The arrows on the photographs are not to call attention to the larger growth, but to mark the years of maximum sunspots. The other trees of that group do not show quite so perfect rhythm as do the marked radii shown, but are like the other parts of these sections, showing strongly a majority of the maxima. Taking the group as a whole, the agreement is highly conspicuous, and the maximum growth comes within 0.6 year of the sunspot maximum. The Eberswalde curves arranged in two groups and compared with the sunspot curve were shown in figure 9, page 38.
In the group of six sections from south Sweden, which were measured subsequently in Stockholm, a spruce (Picea excelsa) was discovered which shows the sunspot rhythm with the same striking clarity as the
| DOUGLASS PLATE 8 | |
| A | 
|
| B | 
|
A. Section of Scotch pine from Eberswalde, Prussia, showing solar rhythm.
B. Another section from the same forest, showing same rhythm.
Fig. 22.—Sunspot numbers and annual rings in spruce tree from south Sweden.
Fig. 23.—Six European groups, standardized and smoothed.
best Eberswalde sections. In view of the as yet unsuccessful efforts to obtain a photograph of this section, its measures have been plotted and are found in figure 22 with the sunspot curve for comparison. In the figure the upper curve gives the actual measures with the standardizing line drawn through them. The middle curve shows the same measures reduced to percentage departures from the line and smoothed by Hann's formula. The lowest curve shows the corresponding sunspot numbers. It would be highly interesting to know the exact conditions under which a tree produced such a curve of growth as this. In the opinion of the writer, it would not be impossible to find other trees of this type, and even to identify them without real injury to the tree, so that surrounding conditions could be studied.
The European groups. — For better comparison, the nine European groups have been corrected for change of growth-rate with age, reduced to percentages of their own means, smoothed by Hann's formula, and plotted in figures 23 and 24 together with the sunspot curve. They do not all follow the sunspot numbers with equal accuracy, and the six groups showing best agreement are segregated in the first of the two figures. The north German and south Sweden groups around the Baltic Sea are the most satisfactory; the group from the west coast of Norway is almost as good. Then come the Dalarne, Christiania, and south of England groups. These six in figure 23 have the times of sunspot maxima indicated by broken lines carried straight upward from the sunspot curve at the bottom. Of the other three groups, the trees from the inner coast of Norway as a whole appear to show a reversed cycle, probably because they were in deep inland valleys, while the southern groups, northwest Austria, and southern Bavaria close to the Alps have combined agreement and disagreement, so that they can not as yet be considered to give a definite result. They are shown by themselves in figure 24.
However, in the 6 groups representing the triangle between England, northern Germany, and the lower Scandinavian peninsula, a variation in growth since 1820 showing pronounced agreement with the sunspot curve is unmistakable. Every sunspot maximum and minimum since that date appears in the trees with an average variation of 20 per cent. This is shown in figure 25, which contains the mean of the 57 trees of the six groups, with the sunspot curve placed below for comparison. The agreement is at once evident. The apparent increase of tree-growth with increase in the number of sunspots becomes still more striking when the means are summated in a period of 11.4 years, as shown in the lower part of the figure.
A second important feature of figure 25 is that five of the eight minima show a small and brief increase in tree-growth. This suggestion of a second maximum is of interest, because in it we find agreement with Hann and Hellmann in their studies of European rainfall and sunspots, and it lends added weight to results which each author obtained but which neither allowed himself to regard as conclusive. In the immense work of Hellmann (1906) upon the rainfall of the North German drainage area, it is this inconspicuous maximum which he finds the more important of the two.
Fig. 24.—Three European groups, standardized and smoothed.
Fig. 25.—Comparison between 57 north Europe trees (smoothed) and sunspot numbers. The trees are from England, Norway, Sweden, and north Germany.
Fig. 26.—Dates of large rings in 80 European trees compared with sunspot curves. Ordinates give number of trees in total of 80 showing maxima in respective years.
Fig. 27.—Tree-growth at Windsor, Vermont, showing measures uncorrected; same standardized and smoothed, and sunspot numbers displaced 3 years to left.
Fig. 28.—Smoothed quarterly rainfall (upper curve), sunspot numbers (center), and tree growth (lower) at Windsor, Vermont, 1835 to 1912.
80 sections. The more recent dates show higher crests because there are more trees. In the second line is the sunspot curve. The matching of the crests of the two curves is unmistakable. The secondary tree-crest at sunspot minimum is very regular, as would be expected from the inclusion of the three groups of figure 24, some of which are evident reversals. This test is only qualitative, but seems to the writer to offer substantial support to the quantitative relation shown in figure 25.
Windsor (Vermont) correlation. — An interesting sidelight is thrown on this type of correlation by the American curves from Windsor, Vermont. The original means of 11 trees are given in the upper line of figure 27. In the middle line these are smoothed by Hann's formula and in the lower line is the sunspot curve, displaced three years to the left in the portions since 1810 and one year in the same direction before that date. The tree crests anticipated the solar crests by three years when the trees were large and making good growth, but when small this anticipa-
Fig. 29.—Correlation curves of solar cycle, rainfall, and tree-growth at Windsor, Vermont, 1835-1912.
If the sunspots are an index of some solar activity so far reaching as to affect our climate and vegetation, it is well to note very briefly their appearance and the suggested causes of their periodic character.
Appearance — At first view sunspots are small black areas appearing from time to time on the sun. In actual size they vary from a few hundred miles in diameter to more than a hundred thousand. Rarely seen by the naked eye, the vast majority are only discovered through the telescope; hence it was only after the invention of that instrument that records of them were kept and their nature investigated. As Hale (1908) has found, they are cooling places; they merely look black by contrast with their more intensely bright background. His remarkable photographs show that they often have a rotation about their own center. They usually come in groups between latitude 5° and 25° in each hemisphere of the sun and are almost continuously changing in small details. Their life is usually less than one rotation of the sun.
Schwabe in 1851 announced their periodic character with maxima every 11 years. During sunspot maximum a small telescope will show 5 to 20 spots, but during the minimum one may search for weeks without finding a spot that can be certainly recognized. Records of the numbers of spots were specially collected by Wolf for many years and later by Wolfer of Zurich. At the present day many observatories are taking daily photographs of them. The term relative sunspot number was invented to convey an idea of the average number of spots visible at any one time under favorable circumstances. The number actually counted receives a simple correction for unfavorable weather or small telescope, so that the published numbers shall be as nearly standard as possible.
While the spot appears black and may possibly be sinking into the sun, it is usually attended by intensely bright areas or faculæ and even by prominences which are often violently explosive, ejecting matter hundreds of thousands of miles from the sun's surface. Thus the sunspot maximum indicates increased activity at the surface of the sun, which, according to Abbot (1913 and l9132), is actually sending us increased heat radiation. During the maximum the magnetic condition of the earth is profoundly affected, as evidenced by northern lights, magnetic storms, earth currents, and variations of the earth's magnetic constants. This relation to the earth's magnetism has been recognized from the first discovery of the periodicity of sunspots. But the effect of the change of solar radiation on climate and ordinary weather elements is more obscure. General effects on climatic conditions have been admitted as probable by Penck (1914), but in general the great weight of opinion has been against a traceable effect of solar activity on weather or climate. From the description above it is easily seen that the sunspots are not likely to be in themselves the fundamental solar activity, but rather an index of something else, and possibly a very sensitive index, for the percentage change in spot numbers is hundreds of times as great as the percentage changes in measured radiation between sunspot maxima and minima.
Suggested causes of sunspots. — The cause of sunspots is still a matter of conjecture, and there is no generally accepted hypothesis to explain them. There is analogy to our clouds in that both indicate decreased temperature. In their limitation to certain latitudes they resemble the belts of Jupiter. The belts of Jupiter are roughly the lines of division between the powerful easterly equatorial current and the slower moving zones on either hand; and indeed this has been suggested as an explanation of the particular location in latitude of the sunspots, for there is an increase in speed of rotation of the sun's surface as the equator is approached. Their periodic character is very difficult to explain. Fundamental periodic changes in the body of the sun have been suggested and, in the absence of better explanations, some such statement hazily indicating the direction in which explanation is to be sought, is perhaps the best that we can do. Planetary influence, however, has often been proposed as the cause. The near agreement between the revolution period of Jupiter and the sunspot period has naturally attracted attention. Stratton (1911-1912) has made a very interesting study of the appearance, continuance, and disappearance of spots on portions of the sun facing toward or away from Jupiter and Venus. A few per cent more spots do originate and disappear on the "afternoon" of the side facing Venus than on other longitudes, but he considers the case of physical relationship not proven.
Planetary influence is sought in a theory proposed by W. J. Spillman (1915). In this theory gravitation is assumed to be due to pressure variations in the ether arising from electronic rotation in the attracting body. The varying speed of a planet in its orbit between perihelion and aphelion, involving varying quantities of energy, requires, he says, an interchange between the kinetic energy of the plant and the atomic energy of the central attracting body. This atomic energy is in the vibrations of the electrons, but he thinks it is likely to affect both the temperature and the electric activity of the central body. The effect in this way of Jupiter and Saturn would exceed the sum of all the other planets combined and is therefore the only one considered. The effect of Jupiter with its substantial variations in distance between perihelion and aphelion predominates, and we have a marked resemblance between the sunspot curves since 1770 and the differential planetary effect. One notices that this interchange of energy would presumably affect all parts of the sun alike and that therefore we could not expect an excess of sunspots on the side facing Jupiter. H. H. Turner (1913; cf. Sampson, 1914) has worked out an hypothesis which is stimulating, even if not yet acceptable. He supposes that the Leonid swarm of meteors, revolving once in about 33 years in a very eccentric orbit, is at the basis of the sunspot recurrence. These meteors were observed in countless swarms, filling the sky for a few nights in November 1799, and again in 1833 and 1866. In 1899 they were expected, but failed to appear in large numbers, having probably been swerved to one side through the attraction of some planet. Turner finds that they have passed near Saturn several times in the last 2,000 years. At some of these encounters a quantity of meteors may have been detached and losing their own velocity may have fallen nearly straight toward the sun, grazing its outer surface in their circuit at a velocity of 400 miles per second, then swinging out to aphelion near their place of encounter, and completing their revolution in about 11 years. Successive returns of the main Leonid swarm, approaches of Saturn, and perhaps even the influence of other planets would be sufficient to perturb this meteoric swarm and cause the variations in period observed. On their terrific flight close to the sun many would be caught in the sun's outer atmosphere, thus in some way causing sunspots.
This hypothesis attempts to explain the period and its irregularities, including the double and triple period. I refer to it at some length because the investigation of trees gives evidence not only of climatic variations in the sunspot period, but of double and triple sunspot periods and possibly of still larger fluctuations. Turner's hypothesis warrants further discussion to explain why the spots appear in sub-tropical latitudes but not at the solar equator. In the planetesimal hypothesis of Chamberlin and Moulton, the rotation of the sun on its axis is attributed to the material falling back upon it after receiving a slight orbital motion from the visiting star. The authors state that the process may still be going on. This view is sustained by arguments based on the zodiacal light and on meteors, both of which seem best explained as planetesimal matter not yet returned to the solar mass. Matter as yet unabsorbed would very likely consist of particles which had been given just enough orbital motion to escape the surface of the sun on their periodic return. The particles for the most part would then have extremely eccentric orbits and pass close to the sun's surface at tremendous velocity. They would be moving largely in the plane of the solar system and consequently would pass close to the sun's equator. If finally caught in the sun's atmosphere, friction would reduce their motion, turning a large part of it into heat and a part into forward movement of the sun's atmosphere. Thus the planetesimal hypothesis explains the equatorial acceleration. A large meteoric group, as suggested by Turner, is therefore consistent with the hypothesis. The undefined zone between the accelerated equator and the slower-moving latitudes on each side would present much mechanical disturbance and favor the formation of local vortices. Such a process as this would be accompanied by the increased radiation in sunspot maximum which has been observed. If this hypothesis has a basis of fact, it is probable that the increased radiation at that time would come from the sun's equator, where there are no spots. Increased rotational movement of the equatorial zone at the sunspot maximum should be susceptible of observation by spectroscopic means. The meaning of the slow movement of this spot-forming zone toward the equator, as sunspot maximum changes to minimum, is not clear under this hypothesis; nor does one see why the secondary spot described by Hale (1919) should have its definite location following the principal spot, nor why the magnetic polarity of spots changed near the last sunspot minimum. These phenomena, recently observed by Hale and his collaborators, point toward causes within the sun.
Length of the sunspot period. — For many years Newcomb's figure of 11.13 years has been commonly quoted. However, recently some of the best authorities say frankly that it may be anywhere from 11 years to nearly 12 years. Schuster (1898-1906) discussed analytically the best known sunspot numbers, those since 1750. This has been followed by the work of Kimura (1913), and especially Turner (1913) and Michelson (1913). In general, the analyses by Schuster and Kimura, and by Turner in his earlier papers, produce a large number of possible periods of small amplitude. Michelson, however, goes to the other extreme. "Indeed," he says, "it would seem that with the exception of the 11-year period and possibly a very long period (of the order of 100 years) the many periods found by previous investigators are illusory." Turner in his hypothesis referred to above reduces the number to a few, which supply a basis for his reasoning. Michelson had favored a period of about 11.4 years and Turner says that only this 11.4-year period is sensible at the present time.
Tree-growth and solar activity. — The correlation shown in this chapter suggests a possible use of the annual rings of trees in the study of solar activity. There are two lines which such a study might take. An intensive line already mentioned includes the search for wet-climate trees showing the solar rhythm in their growth and the determination of the conditions under which they produce this curve. An extensive line of study is obviously possible also in reconstructing, as far as possible, a history of the sunspot cycle from very old trees. The yellow pines of Arizona give evidence that 500 years ago the cycle was operating very much as now. The sequoias, if correctly interpreted, already carry the history back over 3,000 years, and beyond that fossil trees may stretch the time covered in part at least into millions of years.