(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
See other formats

Full text of "The gases in rocks"


This is a digital copy of a book that was preserved for generations on library shelves before it was carefully scanned by Google as part of a project 

to make the world's books discoverable online. 

It has survived long enough for the copyright to expire and the book to enter the public domain. A public domain book is one that was never subject 

to copyright or whose legal copyright term has expired. Whether a book is in the public domain may vary country to country. Public domain books 

are our gateways to the past, representing a wealth of history, culture and knowledge that's often difficult to discover. 

Marks, notations and other maiginalia present in the original volume will appear in this file - a reminder of this book's long journey from the 

publisher to a library and finally to you. 

Usage guidelines 

Google is proud to partner with libraries to digitize public domain materials and make them widely accessible. Public domain books belong to the 
public and we are merely their custodians. Nevertheless, this work is expensive, so in order to keep providing tliis resource, we liave taken steps to 
prevent abuse by commercial parties, including placing technical restrictions on automated querying. 
We also ask that you: 

+ Make non-commercial use of the files We designed Google Book Search for use by individuals, and we request that you use these files for 
personal, non-commercial purposes. 

+ Refrain fivm automated querying Do not send automated queries of any sort to Google's system: If you are conducting research on machine 
translation, optical character recognition or other areas where access to a large amount of text is helpful, please contact us. We encourage the 
use of public domain materials for these purposes and may be able to help. 

+ Maintain attributionTht GoogXt "watermark" you see on each file is essential for in forming people about this project and helping them find 
additional materials through Google Book Search. Please do not remove it. 

+ Keep it legal Whatever your use, remember that you are responsible for ensuring that what you are doing is legal. Do not assume that just 
because we believe a book is in the public domain for users in the United States, that the work is also in the public domain for users in other 
countries. Whether a book is still in copyright varies from country to country, and we can't offer guidance on whether any specific use of 
any specific book is allowed. Please do not assume that a book's appearance in Google Book Search means it can be used in any manner 
anywhere in the world. Copyright infringement liabili^ can be quite severe. 

About Google Book Search 

Google's mission is to organize the world's information and to make it universally accessible and useful. Google Book Search helps readers 
discover the world's books while helping authors and publishers reach new audiences. You can search through the full text of this book on the web 

at |http: //books .google .com/I 

■-- C.// .< i 






r THE OARiraatK Iott i tpt i om or Washhtotok 


Publication No. 106 


• •••••••••• • 

• • •• •• •*•••••• 

.• • .•.•••• 

rRBss OF T B.'iapraooW*ooi»?AW 

• • • 

• • • 

• I 

• •< 

• • • • 
• • • 

.plT JI^ JJgT j>m A 

"•• • • • • 

• • • •• 


It has been known for a long time from microscopical studies that 
some minerals inclose minute cavities which contain both liquid and gas- 
eous matter. For a much shorter period it has been known that various 
igneous rocks, when exposed to red heat in a vacuum, evolve several times 
their volume of gas of quite variable composition. Since these gases occur 
in proportions entirely different from those of the constituents of the air, 
it has not seemed probable that they were derived directly from our 
present atmosphere, unless the rocks manifest some power of selective 
absorption not now understood. The apparent difficulties involved in this 
conception have suggested that some earlier atmosphere was rich in those 
gases. This involves a hypothesis relative to the changes through which 
the atmosphere has passed, and leads on to a theory of its origin and that 
of the earth itself. An alternative hypothesis regards these gases, not as 
the products absorbed by a molten earth from its surrounding gaseous 
envelope, but as entrapped in the body of the earth during its supposed 
accretion, and hence that they are a source from which accessions to our 
present atmosphere might be derived. 

A study of these gases in the rocks has seemed, therefore, to give 
promise of results of some value to atmospheric problems and, perhaps, to 
those of cosmogony. Because of this, it appeared advisable to determine 
more widely the range and the distribution of these gases, their relations 
to other geologic phenomena, and the states in which the gases, or gas- 
producing substances, exist in the rocks. The desirability of supplementary 
work will become more evident when it is noted that, while a considerable 
number of investigators have analyzed the gases in rocks, as will appear in 
the following historical statement, nearly all have contented themselves 
with a few determinations, and that even a full compilation of all such 
results leaves much to be desired from a geological point of view. 


As early as 1818 the attention of Sir David Brewster was called to the 
subject of inclosed water by the explosion of a crystal of topaz when 
heated to redness ; but his studies were not published until 1826. In the 
mean time, Sir Humphry Davy opened the cavities in a few crystals and 
examined chemically the imprisoned liquid and gas.* Piercing a cavity in 
several cases suddenly caused the inclosed gas-bubble to contract to from 
one-sixth to one-tenth of its original volume. The gas was thought to be 
pure nitrogen. The basaltic rock from the neighborhood of Vicence con- 

»Sir Humphry Davy, Plul. Trans. 1822, Pt. n, pp. 367-376 : Ann. de Chim. et Phys., 
t. 21 (1822), pp. 132-143. 



tained gas (supposedly nitrogen) in a still more rarefied state, as its density 
was 60 to 70 times less than that of the atmospheric air. Upon perforating 
a cavity in a quartz crystal from Dauphin^, an almost perfect vacuum was 
discovered. Davy regarded the rarefied condition of the inclusions in the 
crystals as strong evidence that the waters and gases did not penetrate 
the crystals at ordinary temperatures and pressures. This he believed a 
decisive argument in favor of the Huttonian, or Plutonian, school. How- 
ever, a crystal from Brazil gave a very different result; an immediate 
expansion to a volume 10 to 12 times greater than the original followed the 
opening of a cavity. The composition of the gas was not determined. The 
existence of compressed gas in the same sort of cavities seems adverse to the 
conclusions which Davy based upon his earlier experiments, but he sought 
to explain the difference by supposing the crust to have been formed under 
a pressure more than sufficient to balance the expansion due to the heat. 
Brewster^ attacked the problem by observing the temperature at which 
the inclosed liquid passed over into the gaseous state. A number of tests 
showed this to range from 74^ to 84^ F. When raised to this temperature 
the vacuity always reappeared. Brewster interpreted as follows: 

The existence of a fluid which entirely fillB the cavities of crystalB at a temp»Bture 
varying from 74® to 84® may be hdd as a proof that these crystals were formed at the 
ordinary temperature of the atmosphere. 

For thirty years after Brewster the field was neglected until, in 1868, 
Simmler' reviewed Brewster's work in the light of advancing scientific 
knowledge. Studying the liquid inclusions in quartz, topaz, ameth3rst, 
garnet, and other minerals, he arrived at the conclusion that the power of 
expansion of the liquid in these inclusions showed it to be carbon dioxide. 
Some years later Sorby, continuing the researches along the lines suggested 
by Brewster and Simmler, found that the amount of expansion of liquid 
carbon dioxide from 0^ C. to 30^ C. corresponded closely to that observed 
in the liquid of the sapphires with which he experimented.' In these sap- 
phires it was noted that the liquid disappeared when warmed to approxi- 
mately 30^ G. As the critical temperature for carbon dioxide, above which 
no amount of pressure will condense it to a liquid, is 30.92^ C. (87.7^ F.), 
there remained little room for doubt that the gas was largely carbon 
dioxide. Sorby remarked that this gas '' might have been inclosed, either 
as a highly dilated liquid or as a highly compressed gas; but since the 
other ^ liquid has deposited crystals which dissolve on the application of 
heat, it seems most probable that the water was caught up in a liquid 
state, sometimes, perhaps, holding a considerable amount of carbon dioxide 
in solution as a gas.'' 

In the same year Vogelsang and Geissler' heated quartz crystals and, 
passing an electric spark through the gas thus liberated, examined its 

1 Sir David Brewster, Trans. Roy. Soc. Edinburgh, vol. 10 (1826), pp. 1-41 ; Edin. Jour, 
of Science, vol. 6. pp. 153-156. 

» R. T. Simmler, Pqgg. Ann. , vol. 106 (1858), pp. 460-466. 

»H. C. Sorby and PTJ. Butler, Proc. Roy. Soc., vol. 17 (1869), pp. 291-303. Earlier 
papers by Sorby appeared as follows : Phil. Mag., 4th series, vol. 15, p. 152 ; Quart. Jour. 
Geol. Soc., vol. 14 (1858), p. 453 ; Proc. Roy. &c., vol. 13 (1864), p. 333. 

^SiJine water. 

* Vogelsang and Geissler, Pogg. Ann., vol. 137 (1869), pp. 5^75. 


spectrum, which was found to show the presence of much carbon dioxide, 
together with water and a very weak trace of hydrogen. The presence 
of the hydrogen line, however, the authors were inclined to attribute to 

Further researches upon the critical point of the gas in mineral cavities, 
carried on by Hartley,* yielded results varying from 26° to 34® C. The 
lowering of the temperature he ascribed to the presence of some incon- 
densable gas, perhaps nitrogen, while he believed that the raising of the 
critical point observed in some of the quartz specimens was due to hydro- 
chloric acid. 

Forster* and Hawes' investigated smoky quartz, the former distilling 
from the Tiefengletscher crystals a brown fluid of an empyreumatic odor, 
giving reactions for ammonia and carbonic acid, from which he concluded 
that the coloring matter of smoky quartz was due to a nitrogenous hydro- 
carbon, decomposable by heat; the latter made a microscopic study of the 
liquid carbon dioxide in the bubbles of the cavities. 

Investigations which have opened up a broader field were begun by 
Graham^ in 1867 upon the Lenarto meteoric iron. By submitting a strip 
of the iron to a red heat in a vacuum for 35 minutes he obtained 5.38 cubic 
centimeters of gas from 5.78 cubic centimeters of the metal. Heated for an 
additional 100 minutes, there were evolved 9.52 cubic centimeters of gas 
having the following composition: H2, 85.68; 00,4.46; C02,none; N2,9.86. 

As this meteorite yielded about three times its volume of gas, and 
since "it has been found difficult, on trial, to impregnate malleable iron 
with more than an equal volume of hydrogen, under the pressure of our 
atmosphere,'' Graham drew the inference that this meteorite came from a 
body having a dense atmosphere of hydrogen gas. By the same process 
Mallet* extracted 3.17 volumes of gas from a Virginia meteorite. His 
results were in accordance with those of Graham: H2, 35.83; CO, 38.33; 
CO2, 9.75; N2, 16.09. 

Wohler* heated to redness some of the metallic granules from the iron 
basalt of Ovifak, Greenland, obtaining more than 100 volumes of gas which 
burned with a bluish flame (mostly carbon monoxide mixed with a little 
of the dioxide). His results, however, were vitiated by having used an 
iron combustion tube. 

Pursuing the method adopted by Graham and Mallet, A. W. Wright ^ 
conducted a series of experiments on meteorites, which have remained to 
the present day the source of most of our knowledge of the gas content of 
these interesting bodies. Wright's chief contribution lies in his two tables 
showing that there is a marked difference between the gas contents of the 
iron and stony types of meteorites; for while, in the former, hydrogen is 

* W. N. Hartley, Jour. Chem. Soc. (1876), vol. 2, pp. 237-250. 
*A. Foreter, Pogg. Ann., 143 (1871), pp. 173-194. 
»G. W. Hawes, Aioi. Jour. Science, vol. 21 (1881), pp. 203-209. 
*TboB. Graham, Proc. Roy. Soc., vol. 16 (1867), p. 602. 
•J. W. Mallet. Plroc. Roy. Soc, vol. 20 (1872), pp. 366-370. 
•F. WdWer, Fogg. Ann., 146 (1872), pp. 297-302. 

» A. W. Wnght, Am. Jour. Science, vol. 9 (1876), pp. 294-302 and 469-460 ; vol. 11 
(1876), pp. 263-282; vol. 12 (1876), pp. 166-176. 



the most abundant gas, carbon dioxide is the most characteristic con- 
stituent of the latter. His analyses are given in table 1. 

Table 1. 


Iron meteorites : 

Tazewell County, Tenn 

Shingle Springs, Gal 

Arva, Hungary 


Dickson County, Tenn. 
Stony meteorites : 

Guernsey, Ohio 

Pultusk, roland 

Pamallee, India 

Weston, Conn 

Iowa County, la 



















This table shows that in the iron meteorites carbon dioxide in no case 
constituted more than 15 per cent of the gas evolved, while in every case 
but one the quantity of carbon monoxide was considerably greater. In 
the stony meteorites carbon monoxide is low, while carbon dioxide is, in 
the majority of analyses, much the most abundant gas. Hydrogen is more 
important in the iron meteorites than in the stony. 

The same experimenter determined also the gases given off by the same 

meteorite at different temperatures. His figures for the Iowa County 

meteorite are shown in table 2. 

Table 2. 



At 250°. 

red heat 

At low 
red heat 

At full 

Carbon dioxide 














Carbon monoxide 









The progressive decrease in the percentage of carbon dioxide and the 
corresponding increase of hydrogen with the elevation of the temperature 
are striking. His inquiries into other phases of the problem will be deferred 
until the discussion of principles, where it will be possible to treat each 
factor to better advantage, in its proper relation to the whole subject. 

Several years later Wright applied his method of gas extraction and 
reliable quantitative analysis to the gases in smoky quartz/ which here- 
tofore had been subjected chiefly to qualitative microscopical studies. 
However, only one determination was made — that of a crystal from 
Branchville, Connecticut, which yielded a small quantity of gas of the 
following composition: Carbon dioxide, 98.33; nitrogen, 1.67; hydrogen 
sulphide, sulphur dioxide, ammonia, fluorine, and chlorine, trace. 

^Wright, Am. Jour. Science, vol. 21 (1881), pp. 209-216. 


Wright regarded the fluorine and chlorine as being combined, and the 
ammonia as probably existing together with some of the carbon dioxide in 
the form of ammonium carbonate. The amount of water obtained, calcu- 
lated as vapor, was slightly more than twice the volume of carbon dioxide. 

Following Wright, Sir James Dewar,* in collaboration with Mr. Ansdell, 
made several more analyses of the meteoritic gases, and then, in an 
endeavor to discover the source and significance of these gases, directed a 
series of experiments upon the theory that graphite might be the retentive 
or generative constituent. Their analyses of the gases from graphites and 
from the matrix from which graphites have come revealed moderately 
high volumes. ^^^ 3^ 









Celestial graphite 

Borrodale graphite 

Siberian graphite 

Cevlon flnraDiiite 




• • • • 

















Unknown graphite 



Because the quantity of gas yielded by these specimens of graphite was 
so considerable, Dewar proceeded to ascertain whether graphite could 
absorb the different gases when allowed to stand in each of them for 12 
hours. His experiments with the celestial graphite which had previously 
been deprived of its gases indicated that little or no absorption had taken 
place. "It is therefore evident," sa3rs Dewar, "that the large quantities 
of gas occluded in celestial meteorites can not be explained by any special 
absorptive power of this variety of carbon." Attempts to split up the 
hydrogen-producing compound with strong nitric acid and also to wash 
out, with ether, the possible carbonaceous source of the methane, appeared 
to show that the hydrogen existed in a very stable compound, and that, 
while the ether lessened the quantity of methane which the graphite after- 
wards furnished, it did not dissolve out all the carbonaceous compounds 
present, or else that the marsh-gas was subsequently formed during the 
heating of the material. 

Dewar's analyses of gases from stony meteorites, which are in accord 
with Wright's results, are given in table 4. 

Table 4. 











Dhurmsala, India 










• • • • 






Pomice stone 

An analysis of the gas extracted from the Orgueil meteorite revealed 
much sulphur dioxide, which Professor Dewar believed to have been derived 

1 Dewar and Ansdell, Ploe. Roy. Inst., vol. 11 (1884-1886), p. 332 and pp. 541-552. 


from the decomposition of sulphate of iron. In all, 57.87 volumes were 
obtained: CO2, 12.77; CO, 1.96; CH4, 1.50; N2, 0.56; SO2, 83. Leaving 
out the SO2, 9.8 volumes remain, as follows: CO2, 76.05; CO, 11.67; CH4, 
8.93; N2, 3.33. This analysis, the most remarkable of the series, though 
Dewar does not mention the fact, shows a complete absence of hydrogen 
(an uncommon phenomenon), while the percentage of marsh-gas is unusu- 
ally high. There is, however, reason to suspect that there was hydrogen 
liberated, but that it was oxidized to water by the action of the iron com- 
pound, following the decomposition of the sulphate. 

In 1888 W. F. Hillebrand^ discovered that the mineral uraninite when 
treated with acids slowly disengaged bubbles of gas. As the result of a 
well-selected series of tests this appeared, in the light of the chemical 
knowledge of that day, to be nitrogen. Trials with different varieties of 
the mineral revealed a rather significant relation between the percentage of 
uranyl and this gas. The greater the amount of the oxide, the more gas 

Several years later. Sir William Ramsay's scepticism was aroused when 
his attention was called to the paper by Hillebrand, for he hesitated to be- 
lieve that free nitrogen could be produced by treating any substance with 
sulphuric acid. To test the case, he decomposed cleveite with this acid, 
obtaining little nitrogen, but some 20 cubic centimeters of argon, which 
the spectroscope showed to be mixed with some other gas.' A brilliant 
yellow line which appeared in this spectrum coincided exactly with D3, the 
so-called " helium " line, first discovered in the spectrum of the chromo- 
sphere of the sun by Sir Norman Lockyer in 1868. This was the first real 
acquaintance with helium, until then known only as a h3rpothetical sub- 
stance existing in the sun. Lockyer immediately became interested in this 
discovery of helium in a terrestrial mineral, and attempted to prove that 
it was not a single gas, but a compound or a mixture of gases, basing his 
contention upon various strange lines in the spectrum.' 

Ramsay, continuing his study of the gas from cleveite, perceived what 
had been previously overlooked, namely, that hydrogen generally was more 
abundant than helium — in one case amounting to 80 per cent of the total 
gas. The hypothesis that this hydrogen might have been formed by the 
breaking up of an unstable hydride, the form in which Ramsay thought 
the helium should be evolved, if it were derived from combination with the 
uranium or yttrium of the mineral, was put to the test, with the result 
that the evidence pointed strongly against the theory.* A series of miner- 
als powdered and fused with potassium acid sulphate were found to yield 
gas, sometimes helium,* but oftener hydrogen and the oxides of carbon.* 

In 1896 W. A. Tilden made an attempt to determine the condition in 
which helium and the associated gases exist in minerals. Argon and helium 
were of particular interest, for Tilden believed that these two elements will 

» W. F. HiUebrand, BuU. 78, U. 8. G. S., pp. 43-79. 

* Sir William Ramsay, Proc. Rojr. Soc., vol. 68 (189 , , ,^ 

' Sir J. N. Lockyer, a series of six short papers in Proc. Roy. Soc., vols. 58, 59, and 60. 

68 (1896), pp. 66^7. 

* Ramsay, Procr Roy. Soc, vol. 68, pp. fil-^9. 

* Ramsay, Proc. Roy. Soc., vol. 69, j>p. 325-330. 

* Ramsay and Travers, Proc. Roy. Sens., vol. 60, pp. 442-448. 



not be found to enter into combination at such temperatures as are ordi- 
narily attainable. In his own words: 

It also appears improbable that in the minerals from which the mixture of gases con- 
taining helium has been extracted this element exists in a state of ordinary chemical com- 
bination, for, on treating the mineral with acids, no compound of heb'um with hydrogen has 
yet been observed, and the components of the minerals are of a kind which are commonly 
regarded as chemically saturated.^ 

The minerals monazite and cleveite were found to yield gas at low tem- 
peratures (60° and 110°, respectively), carbon dioxide appearing first. The 
monazite heated to 130° to 140° gave gas which, for the first time, showed 
the D3 line, indicating the presence of helium. Between 140° and 250° 
there was obtained carbon dioxide with about one-fourth of its volume of 
a gas rich in helium. At higher temperatures up to 446° (boiling sulphur) 
there was less gas evolved. Cleveite behaved in a similar way. Studies 
on the absorption of helium by cleveite demonstrated that the mineral 
does not absorb this gas at the ordinary pressure, although placed in a 
helium atmosphere for nine weeks. But under pressure of 2.5 and 7 atmos- 
pheres, Tilden believed that he obtained an appreciable absorption. A 
trial with the Peterhead granite, which contained no helium in the first 
place, proved that the granite would absorb none of the gas whatever, 
even aided by a pressure of 7 atmospheres. 

The finding of hydrogen as well as carbon dioxide in this Peterhead 
granite^ led Tilden to investigate the gases inclosed in crystalline rocks.' 
His five complete analyses are as given in table 5. 

Tablk 6. 


Granite, 8kye 

Gabbro, liwd 

Fyroxene ^eiss, Ceylon 
GneisB, Senngapatam . . , 
Basalt, Antrim 











In addition to these analyses, 25 carbon-dioxide determinations were 

Tilden believed the gas to be " wholly inclosed in cavities which are 
visible in thin sections of the rock when viewed under the microscope. 
* * * To account for the large proportion of hydrogen and carbon 
dioxide in these gases, it is only necessary to suppose that the rock inclos- 
ing them was crystallized in an atmosphere rich in carbon dioxide and 
steam, which had been, or were at the same time, in contact with some 
easily oxidizable substance, at a moderately high temperature. Of the 
substances capable of so acting, carbon, a metal, or a protoxide of a metal 
present themselves as the most probable." Hydrogen and carbon monox- 

» W. A. Tilden, Proc. Roy. Soc., vol. 69 (1896), p. 218. 

' Wright's two analyses, showing that trap rodcs yield much the same gases as meteor- 
ites, also served to call attention to this field for investigation. 
•Tikien, Chem. News, vol. 76 (1897), pp. 169-170. 



ide might then be produced by the reducing action of metallic iron or fer- 
rous oxide upon steam and carbon dioxide at high temperature, according 
to the reactions — 

3Fe + 4H2O = FesOi + 4H« 

3Fe + 4CO2 = FeaOi + 4C0 

The origin of the marsh-gas is assigned in this paper to the action of 
water at high temperature upon metallic carbides, or similar compounds, 
in the earth's interior, as suggested by Mendel6ef * and the more recent 
studies of Moissan.^ 

A year after the publication of Tilden's article, criticism of his paper, 
and in fact of the work of all previous investigators in this line, was made 
by M. W. Travers, who undertook to prove that the different gases, not 
excluding even argon and heUum, did not exist in the gaseous state in min- 
erals, but were formed by chemical interaction between the non-gaseous 
materials in the combustion-tube.' The key to his position lay in the two 
reversible reactions — 

3FeO + H2O = FejOi + H2 

BFeO + CO2 = FesOi + CO 

His table revealed a certain relation between the hydrogen and carbon 
monoxide produced, and the quantity of ferrous oxide and water present 
in the mineral. It is shown in table 6. The figures for FeO and H2O refer 
to the percentages in the rock; the gases are expressed in cubic centimeters 
per gram of rock. 

Table 6. 


Chlorite, Moravia 

Serpentine. Zennatt. . 
Gabbro, Isle of 8kye . 
Mica, Westchester, Pa. 
Foliated talc, Tyrol . . 
Feldspar, Petern^ui. . 























• • • • 





• • • • 





• • • • 



Four of these (including the mica of meta-sedimentary origin) were 
secondary minerals whose gas may have been produced entirely by chem- 
ical reactions in the tube, without having very great bearing upon the 
problem of the gas-content of primary minerals and rocks which have not 
undergone extensive weathering and alteration. The only rock tried, the 
gabbro, may be pointed out as unique in yielding only hydrogen without 
either of the oxides of carbon or nitrogen. 

Armand Gautier/ in 1901, came to the conclusion that the gases which 
he obtained from several igneous rocks did not escape from inclusions, for 
the most part, but were products of chemical reactions at raised tempera- 
tures. A small quantity of gas was obtained by heating granite powder, 
moistened with pure water, up to 300° in a vacuum. By heating the same 

' Mendel^f, Prin. of Chem., transl. of £[amensky and Greenaway, vol. 1, pp. 364-365. 



granite powder together with a mixture of two parts of sirupy phosphoric 

acid and one part of water to only 100^, he received more than 10 times as 

much gas as was evolved at 300^ without the acid, or about 1.5 volumes. 

Table 7 comprises Gautier's analyses of the gases expelled at red heat. 

Table 7. 


Granite, Vire I 

Granite, Vire II 

Granite, Vire III 

Granitoid porphyry, Esterel 

Ophite, Villeiianqae I 

Ophite, Yillefranqae II 

(n>hite, Villefranqne III. . . . 
Lherzolite, Lherz 









































1 Av. 





















The r61e played by these gases in vulcanism, and their connection with 
thermal waters, is discussed in more recent papers.^ Following Gautier, 
Huttner^ showed by a series of experiments that when a stream of dry car- 
bon dioxide is passed over a rock powder at a temperature of 800^, carbon 
monoxide results, owing to a reduction of the dioxide, as this investigator 
believes, by some of the hydrogen given ofif from the rock. As the miner- 
als orthite and gadolinite yielded no carbon monoxide, though abundant 
hydrogen, when gelatinized in hydrochloric acid, he came to the conclusion 
that this gas does not exist in rocks. 

In 1905 there appeared a paper by Albert Brun,* ''Quelques Recherches 
sur le Volcanisme," based upon studies of lavas from Vesuvius, Stromboli, 
and other Mediterranean volcanoes. While no complete gas analyses were 
undertaken, much experimentation was done, covering the expulsion of 
gases and vapors at or near the fusion point of the lavas. This author ex- 
presses the opinion that it is the liquefaction of the rock which produces 
the gases, these being engendered by chemical bodies contained within the 
lava itselif. The gases recognized are nitrogen and its derivative ammonia, 
chlorine with derived hydrochloric acid, and hydrocarbons. The nitrogen 
is assigned to nitrides, and the ammonia to reactions between nitrides and 
hydrocarbons, while a dissociation of chlorides furnishes free chlorine which 
may take hydrogen from hydrocarbons to form hydrochloric acid. A 
source for hydrogen and carbon dioxide is recognized in hydrocarbon com- 
pounds, though Brun was less interested in the gases expelled below the 
melting-point of the lava. 

^Gautier, Comptes Rendus, vol. 132, pp. 740-746 and 932-938; Economic Geology, 
vol. 1 (1906), pp. 688-697. 

>K. Hattner, Zeitschrift fOr Anorip. Cbem., 43 (1905), pp. 8-13. 
'A. Brun, Archives des Sciences pnys. et naturelles, Geneve, 1905. 



To obtain the gases for these investigations, the general methods of 
Graham, Mallet, and Wright were adopted, though the details of the appa- 
ratus were modified in many particulars. The gas is extracted from the 
rock material which has been finely pulverized, by heating the powder in a 
vacuum. For this purpose an apparatus consisting of a combustion-tube 
connected with a mercury-pump capable of producing and maintaining a 
vacuum of a fraction of a millimeter pressure is required. Simplicity bmng 
desirable in order to insure the uniform working of the pump in the pres- 
ence of corrosive gases, such as hydrogen sulphide and sulphur dioxide, 
which attack and befoul the mercury, the most elementary tjrpe of Topler* 
pump was used in these experiments. 

To the receiving end of the pump a long, horizontal, calcium chloride 
drying-tube is fused. The ideal method would be to seal the combustion- 
tube containing the rock powder directly to the free end of this drying- 
tube. But inasmuch as the pump and drying-tube are both constructed 
of soft glass, whereas the tube in which the high-temperature combustions 
are to be made must, of necessity, consist of the most refractory glass, which 
can not be readily united to the fusible glass, one break in the system is 
imavoidable. This is made at the end of the drying-tube, which is groimd 
so as to receive a tightly fitting hollow stopper of the same hard, blue 
Jena composition tubing as the combustion-tube. A 5-millimeter tube of 
blue Jena glass joins the combustion-tube to the stopper, and is taken of 
sufficient length to allow of repeated cutting and resealing to successive 
tubes, as they become useless from slow deformation imder the combined 
influence of high temperature and vacuum. 

The capillary exhaust-tube of the pump, dipping under mercury in 
a trough, is bent upward at its lower extremity, so as to deliver the gas 
expelled from the pump directly into the receptacle designed for holding 
it. For this purpose, a separatory fimnel of about 125 cubic centimeters 
capacity, held by a clamp in an inverted position over the mercury trough, 
proved most serviceable. 

In making an analysis, the rock specimen is first reduced to a powder 
of sufficient fineness to pass through a sieve of 30 meshes to the inch. A 
portion of this powder, roughly estimated to approach the maximum 
quantity which can with safety be placed in the combustion-tube, is then 
weighed and carefully poured into the tube through the hollow stopper, 
which, on account of its shape, serves as a funnel. Because the rock-dust 
in falling becomes somewhat packed, the tube must afterwards be held in 
a horizontal position, and gently shaken or tapped, to establish a free pas- 
sageway for the gases, extending the entire length of the tube; otherwise, 
upon attempting to exhaust, preparatory to heating, the air entrapped 
in the powder, having no avenue of ready escape, will expand so rapidly 
as to force some of the material into the drying-tube. 

Thus carefully filled, the tube is placed in the combustion-furnace, 
which stands upon a table of height such that the stopper end of the com- 

^ DeBcribed by Travers, A study of gases, pp. 5-10. 


bustion-tube meets approximately the ground end of the calcium chloride 
tube coming from the pump. The pump itself, installed upon a specially 
constructed table resting on jacks, can be raised or lowered, or tilted at a 
slight angle in any direction necessary to enable the stopper protruding 
from the furnace to fit exactly into the drying-tube. As the whole appa- 
ratus is now rigid glass from end to end, care is required in fitting the two 
parts together, lest there be strain suiScient to cause serious fracture. 
To prevent leakage during the extraction of the gas, the groimd-^ass 
connection (the only source of leakage) is completely incased in a thick 
coating of parafl^e.^ 

The air in the apparatus is now pumped ofiF imtil the exhaustion can 
be carried no further, at which point the pressure may be in the neighbor- 
hood of 0.01 millimeter. If allowed to stand for several days this vacuum 
remains entirely without change. When ready, the burners in the furnace 
are lighted, the separating funnel in which the gases are to be collected is 
filled with mercury, and the evolution of the gas is under way. As fast as 
the gases are liberated by the heat they are pumped over into the collect- 
ing-funnel — a process usually requiring about 3 or 4 hours before the last 
traces of gas have been expelled. 


After constant temperature has been established in the room, the gas 
is drawn from the receiver into a Lunge nitrometer and the carbon dioxide 
and hydrogen sulphide absorbed by the introduction of a cubic centimeter 
of 30 per cent potassium hydroxide solution. The remaining gas is trans- 
ferred to a gas-burette, filled with water, after which the remainder of the 
analysis is carried on according to the method described by Hempel.' 
From the potassium hydroxide solution the amount of hydrogen sulphide 
absorbed is determined by titration with N/100 iodine solution. If it be 
desired to measure the quantity of helium and argon, the gas remaining 
after the removal of all the constituents, except nitrogen and these inert 
gases, is passed over metallic calcium heated to redness.* This absorbs the 
nitrogen, leaving only helium and argon, which are examined spectro- 

^ Whenever rocks containixig a large proportion of quarts are teeted, it is neoessarv 
to substitute a porcelain tube, since (luarts scratches glass, causing it to crack when hMted. 
The connections are readily made tight with paraflfine. 

' Hempel, Methods of gas auafysis : Technical method. 

* Travers, A study of gases, p. 102. 



The analyses are numbered in table 8 in the order in which they were 
made, and therefore furnish a chronologic&l account of these investiga- 
tions. At the commencement of these studies, it being deemed advisable 
to make a rapid survey of the field and establish the range of the phenome- 
non, in order to direct later experiments more intelligently, less attention 
was given to securing the most trustworthy method of extracting the gas. 
Twenty-two analyses were made during the preliminary trial stage, before the 
apparatus was overhauled and sealed glass connections substituted for nibljer 
tubing. As it was difficult to prevent a slight leakage of air into the tubes of 
the ori^nal apparatus, the first 22 analyses are characterized by a higher 
percentage of nitrogen than those made afterwards under more favorable 
conditions. Hydrogen sulphide was not determined in the first 17 analyses. 

Table S gives the percentage of each gas in the total volume of gas; and 
the volume of each gas (at 0° and 760 mm. pressure) per unit-volume of rock. 
Taslb 8. 








Sperlmini Ko. sod nmub. 

















1. Meafum-emlnca vfhile BreniU!. mUier 








low i» quaru. 



[pldgHr phenocrygla. 
B. KMW»5n fcbia from Mcs.bi diBlriot, 






31 .M 



Minn. Fn>m oldtrt known scrCea ol 




LsliB Buperioc iMion, hlehly tarboii- 
ated. (SiKTiraen 108%. sTide IWfifi, U. 

8- G. 8.) Prom V. B, Van UIpb, 

P. c(. 







Gnelii Irom cod lact of Kcemtln greeo- 



Honeuid lAurentlan KnuilCcDurRat 

ftirtKB, Out Fram C, R. Van BiK. 





rram acuart Weller. 



B. Fortin ihalo. Iths™. N. Y. Prom Dr. 
WSlcr. INKroKPii omltWd froio analy- 









it* becmnn of I«ikBKc of air Into tube.) 

7. IIIUOOTlte, Canada^ nmltiial taken from 

P. ct. 


klMKC dab ol mica lu Walkcc Uiueum 
8. Poumam mndstone, Bjusboo.WlB. Treat- 










fl. Pik™"po^™K™iIiw, f™ ni^We-road 

P. et 







Eiained. yullowJuh-pink gniifte cdhbII- 
8ao?-««lDwl™lnkwr ■"iSiUo o" the 

peak proper has bteii inlraded. 
10. Rhyollle. Uarble Mounlsin on north 
iilOT«« of San Fnuu'lsni Psak*, Arii. 










11. Keweenawan dfabaw, Irom 1 milp oM 

P. Cl. 






at IJriwer Junpilon, Foik Co., Wis. (a 



12. ftuarliilc scbiil, Follpfted 2 mileawralh- 

wanolBaraboo. Wis, 



IS. AndeRllu. silver (,Tcek Ba«lQ.Oui»s- Co., 

P. CL 






H. Coanw pon^m f"™ dump pUeoI Ban 
Pedro MfncdursT Co., Colo. 









Tabli 8 — Oontjnued. 










U. Oftbbro dlorile Irora Bboul 13,000 tt cle™- 








tloD on souili «ide a! M t, Hoedels. Colo, 








IB. CcsiB&greJaed gabbra dlorlle, Bummlt of 

P. Et, 






•ec. % DolDlh, Hlnn. SpeeUnen 160^ 
BUaB«02,D.B.G.e- ThUtoonenlothe 














Bt Looto RIvra aeria (HMO tt thick) 

wUcb b the lowot member Dl the Ke- 













hMdtodtmWedihale. ' 

"Ui MMdve emnite porphm Horw Race, 













•n na&Kralnod gnelm. buqd rock-mu u 








„ r6.'l; JGESS.g'^.SSiS'"- 







DUVuNm. IDand^. Specimen 11(M. 
SUde 5701, V. a ti. a. CollecUon 0. H. 












Ulch. Bpedmen 14302. aOdiinH. from 










C B. Vui HIM. 

». SchlMirlttacbloillohLBUckHIlIi.e.Dia. 
Docrlbed by Von Hte, Bull. 0. 8. A., 














__ BiienM of bslbolllhlc gianile. 







Obtained Iram BliUie « Oo., nuuble 


































P. CL 









































i; Fc^Oh 8.001 FeO,4.«4jOBO,ej!2;HgO,U«;Na,0,tJS: 

I Carbonated. 

>No.l7. An&lfiila(byBtn»iK): B10b49JS; AliOkll 
E,a 1.81: lW).l-Mi 100.80. ^^ 

•Nal9. Aiuayili(^BLBnILei,p.lU): B10i,M.8S; A1,0» 26.40; re,0).l.tti hO, L6B i CW), 8.W ; HsO.LSS; 

,0.s.ee: KM.i 

m collected fr 

D Inlnnlria of giuilte-gneln lata fcraembme, which hu been odled 

■jupa iiuiuuBn br BiDOk^ the gimnilfrfnelM beliw placed at the lop of the Upper ElUDiilan. BnlletlD 82. p. 2S. 

TtuavrisihowasgmdatloDframmaalTepoiphyTl&Knnlte loTcr; fli]»-baiMle(IaiKlr~ "- "••-• ■■ •— 

o[ the Intmdre dike. 4adnB(teialmoAliItpe—*^' "--"■ -■-■ — '--- 

t™. No.21i««lmllartoNo.aO,boln)tooUl ., 

tieuigmore like ftdU)rlte In comixieltlaa. nM low gea-Tdiime of No. 20 In oomparlnn with the two ot 

mem Is peihape lo be explained by tlu bet that It & a much mora ponxu rock, cmmbUDK (MuUI; to a flne poinlM 

Te poqthyrlilc Kianlte lo rer; floe-buMled apelM. Ho. 19 l> fnm the oenler 
bitp«TOptIbIr on both aide* Into the IbMMTalned snelM of which No. 20 tt the 
DO the ddei, bat to banded. Thto rack & more Baiic than trplcal gianiic, 


«iof CHb^and 

M Utile N, pneent. 


Tablb 8 — Ctmtinued. 











87 JO 






lesIon. apeclmsn luee.SUdaG74f. Db- 









BCrtbed bj Wllltams. Bull, ffit, pp, ea-7fl. 

Ad eilremelr Kllered gabbro, ooolaln- 

ins calclte. teilclte, cblorlU, uid lao- 

Mich. Specimen UTlTWleBSei.D. a 
















O.8. rrom C. R. Van HIM. 
















elcen nil ifIilh i.anvJor nns hesUsL 
i-rom C, B, VBn illsB. 
















































DeKalbOa.,a& Fram & W. UcCalllc. 









IB. Omnc reddish Archeu Kiaalle. BIk 










ottboclaJie form largecry91s.lt: queru 

!■ abundant, but bintlte Is mther spubc. 


lb. OrtonTllle Kmnltc, bintlie crystnln u[ loat 






epwlmen stpanited (rom (luanx oiid 
gravicj' solution. 





















iL Bftmllton ihale, Nemom's euUon, 







IB mile* W. ol IfuhTlUe, Tenn. 
Proni um AugoaU T. Hinlock. 








Kttll; eilned wi&ln the rock In 

tbe Eueooi etate can not be staled : 
mp^trf It probeblj came from de- 

ent Heavy brown tan aim pro- 

149. Oll-rw* from !!Sd-iInc mine n«« 









PlatWrtllcWta. From H.F. Bain. 








Quebec PminP. D. Adanu. 












Fe,0,&jei0i0,UGiUS0,llJ«iHa,0.2M: EALM; 

1K0.S3. All«]nl«:810t,«BJ)6: JU,Ot.3t.7S:raO,B.«5; Fe,0 
HA7.SS; COhO.«; 100.% (R. B. Rlgn BiiU.83,p. 78.) 

) Probably onlf » muUI put of pi wH leallr irannt In the I 
what • bltmntnoui ihale may yield. 

■Ametamorpluaeditotaof tbeOienTtlleieilea. BefeieiiM: Am. Jonr. BcL, rol. EO,p.SS. SeacrlbediaaflDa- 
Bimlned nrnetlferoai allllmarlte gneia containing mneh qoafti and orthoolaae. Graphfte and pvrlte ftlao pnent, 
MUlnB the gnelM to weather to » Very mitr oolor. Analyil* by V. V. Bnnt, ot UoOlIl OnlvenUr : So. U. BtO^ 
aiM;Tlp„!.06; AlfO>,J9.n |_Pe,0| and FeO, LOO ; Fed* Ut ; XnO, tiWM : OaO.OJSi KgO.LSIi ilifi,Q.n;Kfi, 

SelM to weather to a very 
8: AlfOt, 19.731 Pr " "' 
iM 1 BiO Xlgnicko). 1.82 1 WM. 

Table 8— Continued. 

U. Nephelite (yenite, north ol HouDtBln Lake, 
Tp. of Uethuen. Ontulo. F. t>. Adaitu. 
ol Igneona origin. 

4S. Inm-bwrluB buKlt, OvUak, DlMo Iiluid, 


«. lUgnetlte nad Snaka River bed. Idaho. 

trnpuilllea b; magDet. 

47. Andcdle, Bod UonnMio, HW. ol San Fnn- 
daoo PealoL Ariz. OcillMted b; W. W. 
Atirood. lb. Arlliiir Tk^lor dacrlbei 
thlt took u ui andcalte porphRr wbon 
■Toinid-mis {H p. ot. ol ue wbole) oon- 
lliti of 48 p. ct. pjTOzaie. n p. ot pUglo- 
claav (tobnuioctte), and U p. ct mu- 
nKlta. Tha phmooiTiti an llnw^ou 
fddqiH, angfta, bonibleode, and mas- 
nstlte. Oocnn aa Inegnlai blocki ee- 
UMiied In ttn tuff or TOlnnto biMcia ol 
which Red Honntaln la bnllt. 

IS. FnraraiB cr7aUl& Red Ifonntaln, Arii. 
Collected by W. W. Atwood. Tbese 
cryat&ti, noglng In ilie bum ■ beui lo a, 
■mall marble, were louod looae on Ihe 

-. _....iangle, Colo. Spec 

Ho. UH V7&. Q. a. A kv& flow bolong- 
' r to tha Inleimiediale seiica. From 

). KccACUt 

dark^^soined nwk, with tbe 

, —Tlalble to the naked, eye. 

ConUtna UHle or no leldipai and eom- 
jandlTel]' little nebhellne. Olivine and 
•Dills an the mon important romtltii- 
enU. The pnaenee of mellllte IndlcaCea 
'amoiiDtoI Ikoelnmapn*. Age 

k ol Shonkln 

(Ft Bentoo Folio, p. 1} m » nwk at — 
■yeolle familr Terr iloh In auslle, con- 
talntng aooeiMr* oIlTbM and blMk mlea. 
White the chief lJabt«ol(ired aoutltDent 
Ii orthoelaae, nepnellta and aoOallte are 

aga. Collected 

dunbtlDlI} pi 

ceant In TmrrlDK mnnm 
'e of proliabir Eocene 
W. a. Weed. 

I. Thenllte, frotn tbe laccolltee on Upper 
Bhleld River RuId, Craiy Mountafue. 
Mont. U. S. HBtloQBl Museum Ko.TaiSK. 
Collected bf W. H. Weed. Described 
(Uttla Belt UouDtaina Fidlo, p. 4) aa a 
datk-frav taaaltio lock eonunon] j occoi- 
tlDg in abEal*, bat mtlj In dike*. Poi^ 
ph^Mc 1 1 J ilala of anglte fonn the moat 
proainent phenooryM, atoagh large 
plalaa of brown mlea an oommon. Ool- 
oilen pan D( the ground-maM I> a gnnu< 
lar miiton of nepbellte — ' " ■"- 

H. Qnuti lyenlte porphyry, amnmll of En- 
flnaet Honnlaln, Bllverton Quadmiiglc, 
C^ No. ST3S V. S. O. 8. From Dr. 
Deacilbed (SUTBrton Folio, p. 11) 



Table 8— Continued, 

Specimen Ho. and rmatka. 

P. ct 








Eltber an intr^Te or i«(dng apon anda» 

KIdgn. Alteration doe to Intnuloo of 
joggtj^(8e.«i»ly.laNo.8i.) From 

». OviMlleraM inelM. Dwwin'i Faiia, Tp, 
of Hawdeo^ebe^nobablTcrf ie<fi- 

EC nS^Srade nenile. Cape' Oi^betb,' He. 

talni nnmenrai &r£ qnarta crntels and 
dear uldlnea. embedded In alight gny 



P, ct. 


























.„...».. „,™„. 













forks ol Deep Creek on lidge from Mt. 

Boffner, a mifa from lower contact, TeU- 
urldc QoadrauBle, Colo. No. 2SK U. a 

p. cL 

p. ct. 



P, ct. 





























62. Andealle. Kodta Hills, Cutter Co.. Colo.. 
from summit of small hill east at spring 
on ROBiU Road. Dike faciei oC Fifnglo 

from CherUey, Quebec. F. D. Adanu. 
6*. Aodcslte, Uparl Islands. 

Harburn Tp., Ontario. Dr. Adams re- 
gards this as almost cettalnly an altered 
Sediment, lor It occurs In beds inleiBtratl- 

pyille and conlalna giaphlie, Aaalyaia 
06. Gnel« ol igneous origin lot as, R. 9. 
WollasloQ Yp-, OnUrio. F. D. Adams, 

ton Tp, dniario. Probably ol eedimea- 
loty origin. F.D. Adanu. 









K*iO. B.OT : HtO. o.es ; FfOi, ai 


Tabu 8 — Continuml. 

No. uid rfDurki. 

eg. Amphibolic. U&Tirell'i CttMiag, Glamor- 
gKH Tp., OnlHrio. A hlgblr altwed llmc- 
Mone. Dr. Aduiu. 
t. Blcc iwk, CuudMn Pid£c R. R. 0,2S mile 
■ul at 8iidbiU7, Ont&iio. Probkbly im 
kltand NdliaaDt ol Hun>nI*D age. Dr. 

TV. Andedle. QimDlie 1 
Utah. AbKcollteiii 

71. Vdn qoaiU, Oiuille UonnMln. Iron Co, 
VXtiL Aaaodued wltb Iron ore whlcl 
bw dereloped mloag the contact of tbi 
UeeoUtauMlntnideaUmeitoiie. Aoooid 
Ins to Lelth It tni premcubly decided 
fioD the andMlle (Na 70). 
L PbODoIlIetnch^te, cut DUtot Bull Honn- 
UlD, PIke'B rtmk Qnubuigle. Colo. Ka 
ausu.ao.a OcKrlbedbyCnwIFlke'i 
Fenk Folio, p. S) u ■ demtk gnildKTeen 
iDck with tabol&r orxau ol Hnldliie, 
which give it B typical potptyritle t«x- 

74. Dlotllk Dtet Ue fheet. FeoobMOl Bov 
- ■ " No. 11S7. From E. 8. 

. _. , T lanow* of ihe 

Bmboo, AblanuiB, Wli. Hard, tndu- 
nted while nnditone, not lu from Ihe 
contact with the qnaittlls. 
■7$. Patadam BaudMone, Ablemani. WlA 

S. Fennbui red nndnone, OaMen ol the 

Ooda. Mar Oolontdo CIV, Colo. 
). Topu qi — ' ■■ — °-'- — 

X Bt. PaUc ouidatoDe, Minn.h.h. Cieek. be- 
low the lall^ MtDDOipiiU^ Hlim. A Kjfl. 
roDBrtatblT while, coanefnlned nnd- 
HDoe. bAw heallns in the tube the 

I. Bt-PMerandtfone, Ulnneliaha. The sand 
Na SO, pvlmlKd and heated a second 

82. ftoartilW. taloDglni 

urdon Tp., Queb 

Indkateil that tlier «ero fonned from 
aqueota aolatbm. Eatlrely tmniparent 
and wlIboQt Tldble luclaHona. 

- I, Indian TiaU Bldse, U Plata 
'e.0olo.(aabeetlnl' - 

nanandfemle mloenli, aod mora aoclte 
than hornblende : feldapara much Bericl- 
tlied BDd obecuied by chlorite, epldole. 

which WBi DuMe to ocape might not 11111 remain within tti 
ji tbe (reah laud, moit of the gaa In the Snt trial apparently c* 
















too. Hiob. Ftom A, C. lAne. TbSt 
pleoe Game from a depth of G24 feet. mMB- 
^^ alonsa^ 4^^tlui >>°^>^ 

a»elmeD ai Noa. 88 and 87. 



p. Cl. 
































SpHtmeu No. and nmarki. 














9S. PiWhblends. Brnver Co.. Colo, From 

'94. CarnoU'te, Coiomdo, from H. N. McCoy. 
Uranyl vanadate, probably with some 

From C. K, teitb. 
H. OraneHU rook, Mvwbi dMrlpt, Uinii. A 

From Pr. LellH. 
97. Mi<«tKiu» guartiiw. I'lnta MtB.. Utah. 
rrom Dr. t^iih. 
























_ . le from bydrocarboni (oinble In alcohol or etber, thla material, after 

being verr DnelT pulrerlted, waa dlgeated with alcohol (free from organic Imporlttea) for 20 houn : then with tat- 
free etlier IOT4t houit. It wai then thoraughly waihed with ethet on a Alter which bad prerlouily beea treated 
Ih the aame tat-free etber. Afterward! dried at 100° Id aD oveo. 
'To get rid ofallcarbonatea the powder wBi treated with Goncenlntednitrfaaold for 88 httan. Unchgaawaa 

renoB. including a eopiona evolution ol nitric oxide. The powder waewaebed until all ti ■ --■' 

.—. .^ 1....... '-)d In an air-bath at 1180. This malerlal he ~ ' "" "" 

lUrte add (or three days In a vacuum. Thi. , — 
rbid by barium chloride, and then dried In aD oi 

u maintained between tbeee two oombnrtloni. 


Tablb 8 — Continued. 

se. Quartiite, tUb HIU, ccai WBaaan, Wle. 
From ekmuGl WDldmBn. Famoua for 
IB s" babbles. Powdered on sii aa- 
vU umI meuillli: iron Ihui introiluoed 
removed aa auDpletelr u poolble wlib 

199. QiBLrdle, Rib Hill. WIb. game rpeelmen 
uNo. W. GTanules uwil lasuso ol tine 
powder. CnuDbleH rend ily Into grao 111 efl, 
vhicb were tresled wiUi boHlng by- 
dnchlorfai acid Eo iBmove uiy iron 
wblcb mlsht IwTe oome (rom tbeinvil. 

UO. AmUwon^PTi l tllMoninnartt, Cmgo. neM 
OtwueTNAW amtti AloL No.^N. B. 
W.; Jlo.liaiU. o( W. Dr. Ulth. Re- 
dnMd to ■ eouM pcwder on an ■nrll 
■Dd tmtad Willi bolllDg dilate hydro- 
cblorloacld wbleh removed — '— ■- 
' " ' '"it being eareti 
u toorv Hiiely : 

lOL Bay) faom penuUle dike. New EngUnd, 
WUksr HnraiD CoUecUon. UMerlal 
far tlita mnaljvii t&kcn from ■ tnavlve 
^trfl, g bwbM In dluKiter, u tranrau- 
■DtHwliidow.BlBBuid witboat Tlnble 
InfflnJmnoi luumritla. InRtaad ol be- 
liiC pnlTeriMd ae mMetU wu mod In 
tte Km ol amBll tnwmenta wbleh were 
wadiadwtth bolllDgliydroobloilc acid. 
Alter beatfog, the Inaapareut Ing- 

lOIo. Am 7t b^Swied to eipet all the gaa, 
the Taenam wu nulnlalned oTcniigbl, 
and the material healed 4 hours maie 

coDtalnlDK many eryitals 
- "heaterfleld, Uaffi. Wall 


1C8. Qoarti Inno pegmatite vebi, Jonaa Fall). 
Baltbnora, uZ Ka. am, CqIt. ol Wli. 
aoDeotloii. Dr. Leltb. medmen con- 
talned ■ lane eintal ol mlciaolfiie over 
3 tncbea In lenstli. Indicating oondltknii 
favotoble to crntalllialloiL Nomlcro- 
eUne mm used tot the analyah, howcTer. 
II pGomatlte* contain mocb nt. It wai 
tboiiKbl this one ahould yield a good 

IM. Albite, Olbb'i mica mine, Yancey Co., N, 

Ids. Quarta from Miocene lava. Iron Co., Ulah. 
No.*661*U. S.U.S. From E. C. Hanler. 

lot. Allegan meteoiile, a Btony aerolite which 
ItS at Allegan. Hlcta., /ntr 10, ISW, and 
waa dog oat of the aand eaXl hot, within 
t mlnntea of Ita (all. From tJ. 8. 
XaHoul MDamm thraogb Q. F. Hei^ 
ML DtMrlbed bf HcnOl and eiokee 
(?>[iB.WaaliiD>tan Aoademy ol Sdeooea, 
vol. », ML n-a)- Before etttiaatlnK 
lliegM,ine imwdered meteoric material 
wu heated in a vacuam at lJin° 'nr 3 
boon In the presence of phi 



n to enable the dry- 
rb all moliture not 

■TheToIomeof carbon dioxide being much neater In the case of the giannlea Chan In the finely induced 
powder atinngly niggeats that much of this gu 1* mechanically incloaed in caviUe* within the quartiite, and 
OKBiiei when the gnnulea aie putrerited. Tbla nilninn waa itrenglhened by a alight cncking noiw which 
eame fmm within the tube aa aoon u heat wu applied. The gu came o9 wltb a nub wlien the tube wu heated. 

'Ko.108. Chemical analnia by Dr. H. N. Stokes: MetalilcpArt, ,!S.<Mpercent.,ufollowa: Fe,21.0ll; Cii,,Dl; 
in.lA;Co,.U. Btiony part, n.M pet cent., u follow! : SIO,, M.K : TIC^.IH ; PA,. 17; jLWt.2M; Ctfi^ .69: FeO. 
t-tl; r^&M; MnO, .Q; CaO, I.TS; UgO. 2t.W; KtO, .23; Nb,0, .M; IV> at Ol?, JM; above 110°, Jt; lOO.OOv 


Table 8 — Oonoludnd. 












■107. Eitacado metoorlle. fell near Eetacado, 
Tex.,lal8Ki. De«!rlh«lbyK.B.Ho*»rd 
(Aid. Jout. Scl., vol. 2i (l4oe), pp. 65-60). 
Kept iTi K Tacuum at onlluary iewpen- 















houn: then bated U Iso" lor 6 hoon. 
and BUawed to remHn In vmouo (or 19 

(ramTolQW. Men™. 



















i Third detannioBUon 







"109. Irotiorp, Iron Co,, Utah. No, , U, 3 G 6 

kud ■□Iphsle : 

Flnl portion (gu whloh oune off In 







Second portion jgM which cameoff In 

Third portion (material be«W3honni 










5Sl :" 




111). Basaltic IbtB. Rllanea, Hawaii, CaKcade 










oflHW. A rather poiou* U\-», having a 






.a ,S»iS»SS,u„.™. 





of April. 1306, ooUected by F. B, Taylor. 









Uarcb !», IKT. From •oath aide of a 

quarrrintiDntor the ohurch In Boko 

Bpeclinen cune from 10 ft. below the lop 

and ^ IL fram the bottom of Ihe Bow. 

which had been blamed at tbla point. 

'lUL Frenh lava, Vauvliu ume flow »i last: 








0.5 kUomtWj BE, oi church where No. 




Ill was collected. About 8 ft below 

iiiirrace and 2 II from bottom ol bed. 

Collected March 30. 1907, by F. Taylor. 


'No. 107. Chemical analysli by J. U. DaviM>n : re.14.S8; Nl. l.SO; Co. .!«; 8,1.37: P, -15; BIOi, 3&.B1: FeO. 
16.63; Uia 22.74: UiO,3.M; Alk)., S.«0; Na,0, 2.07: K,0, ,32 ; 100.96, 

* HO. 108. Average ol IS anjayaea compiled by Fairiiiglon ( Pub. Field Colombian Uiueam No. 130, pp. 82-84): 
Pe, W.U; Nl.7.90:^, 0.<S; P, oi4: B 0.14; SI, 0.02; HlK., 0.67: W.16. 

'lion bonngiand fUlnnwere uaed, butwftb cheae there waa included a little nut, wblch adhered to themeUl 
when It wai wluidrawn wiih a magnet. 

« The ma,terlal naed In tbla delerminatkin coDBlited of bright borlngi csrctollT (reed frotn rtut > pocket of which 
■—*»■""•-'- -^- —."«'"*"«• 4.. j-iiu— n..f (.. ...i*....#»....t. ■»» —..-'■-^ 1q diawliu oat tike metallic borings 

._ .^ , ,..th themetu In the ccmboMlon tube. 

d to much leas than In the fliM detetmlnatlon. Mo flUnga were uaed. 

iHoa boringi from the inleilor of the apedmen weracairefnlljrnadeby Wm. Gaectoar A Co.. 

Botemiao-hutniment maken. There waa no vUble nut adhering to theae boifnga, which were then woAed twice, 
with a magnet, wlthoat any fmpuritlei bedDg left behind. The white paper on wnkh thli (iteration wae perlonned 
faOedloBhow the illghleat dlaooloiBllan, aach al It bad done In the two pnvloua detennlnatloni. Anuoal the 
material waa healed at 100° for S houn, m the preaence of PfOt. and then allowed to Mand In the —"■••'" M_i-h> 

.. . ,_. — ,«. — » .^. wai perfect at the «id of thlattme, Che flrattw., — 

olvea when heat waa applied, were not kept, dnca 

. at the tnd of thlattme, Che Brat two pnmplngi ol gaa 
. >■— waa applied, were not kept, dnca It wH dstrad 

lranend<Hu dect thepraaenccol a little lion 

Aeompailaon M the , _- , 

nut will have upon the gaota emdved from k metallic meteorite. Much of thla gaa la doubtlen derived 
caibonale and the hydimted oxide of Iron, aa will be explained under the topic of gaa due to cbemlcal 
Oimt care la therefore neceMur in making gaa analjili ol Iron meteorila to avoid an* contamination ol nuk 

• The men atriklng leatnre ol theee analyaea Ii the nnnaual amount of aulphur dioxide, which Indicates ai 
Olldlied condition ol the ore, 

' The odor of anJphur dioxide wia very prominent In Ihe gaa obtained bom then two Veauvlan lavaa. 


As the volumes and relative proportions of the gases found in the fore- 
going analyses vary within wide limits, the nature of this variation can 
best be shown by grouping the results. To make these tables as complete 
as possible, not only the results of the present studies, but all the available 
analyses of other investigators, have been included in the lists. Except 
In the case of four of the five analyses by Tllden, relative to which suffi- 
cient data are not ^ven, all of the figures in these tables refer to volumes 
of gas per volume of rock. Previous investigators have usually given the 
total volume of gas and the percentages of each constituent. From these 
I have calculated the volumes for each individual gaa. 

Tabu e.—AnoJyiM elaasified by gnmpi 









Buck nai loulltj. 

H,S. CO,. 






OraniUiand gneUm r/ ignema orisin. 












































11 w 




OslnbT Hub snnlH, oeorgU 

Blone HoudC^ gTBuIte, OeoiglK 

OnonrUle e»nlte, lUnn 

OmiiitB porphjTT bowlder. New York. . . . 

Average of IS uukljMi , . , 


NephoUle irenile, OnlArio 

abonkln[te. Hlgbwood Uountkloa, Hoot. 












The oalibTOJlionte ffronp. 

8S£S:li?3-^"""' *^ ":!::::::::: 

































Keireenswan ai«b»e, Wtooooiln 


Tasia 9. — Analyf daarified by group* o/ rodcM. — Oontiiuwd. 

























mibtun oRil doHlte-Cont. 






















































AndcHlle, Red Uomttaln, AHl 




AodMlt^Oi^M Mountain. Dt.li 



ATcncool Taoalrm 

Rhjollle, Marble Mountain, ArlMO* . . . 
RbiroUle Tltropbyre. Telluride 































OMim porphyry, Onray Co., Colo. 

Topai qiuirt»porpli]TT. Saxmj 











Smokf qaarti. Btancbvllle, COQQ. 

































Ppaxo^fnelH, Cef km (p.ct.) 



















Table 0. — Ano^Ms doMifUd by gnrnp* of roeka. — Concluded. 


























Avenweof flret3«nalT»M 

SuiieHlut,paiTdBradmnd aiteUMi'.'.'.'. 





















Table 10. — Variou* minerab. 


Hluml and locality. 



m. Bt 




































Klwhln Md 



Chloriu, ZopUn, MonTia 


FcldjpM, FeteihtBdEmiilU .... 













PltrtW«n<l^Be»TBrCo?, a>]i^'.'. 









It should, perhaps, be stated that in making this and other averages 
of aiial3'Be8, in those cases where, on account of excessive carbonation, no 
figures are ^ven for carbon dioxide, the average amount of this gas cal- 
culated from the other analjnses ia assumed to be present in those rocks 
marked "carbonated." This addition is added to the average total and 
makes this figure slightly greater than the average of the column which it 
foots. The same method has been used for carbon monoxide in the three of 
Travers's analyses where carbon monoxide and hydrogen are put together. 

Tabu II.— Sloi^r « 










































onroeu 1 

Allegan, Mfeh 



Avenge of 19 uuljni 

The figures for the Orguell meteorite which yielded euch a remarkable 
amount of aulphur dioxide make the average for the sulphur gases an abnor- 
mal one. The presence of this gas in quantity must mean that the meteor- 
ite has suffered much from weathering and oxidation subsequent to its 
fall. Considerable troilite has passed into iron sulphate which has been 
decomposed by the heat of the combustion-furnace. 

Omitting the sulphur dioxide of this specimen, the average total volume 
of gas from stony meteorites is reduced to 4.80 times the volume of the 
meteoritic material. 

Table 12.— 7 























I'S' "m 






A»ai»ge onjiling Arvii QiBleo'rite 

,67 1 M 

1 Flight, PMl.TtMl». So 

17a ( 






Methane was determined in only two of these analyses. In these two 
it averaged 0.10 volume; but in order to make the figures consistent in the 
table, it was necessary to average these as if the eight other meteorites 
yielded no marsh-gas, though it is highly probable that this gas was present 
and has been included in the figures given for hydrogen. 

The unusual amount of gas from the Arva specimen recalls the be- 
havior of the Toluca meteorite,' which, at the first attempt, produced 24.42 
volumes of gas, owing to the presence of a small quantity of iron rust, but 
whose pure metal evolved only 1.85 volumes. An average, omitting the 
Arva, is therefore made. 

> Ante, p. 22. 



Table 13. — Igneous rocks. 




Type of rock. 

Basic schists 

DIabMes and basalts 

Gabbroe and diorites 

Oranitet and gneisses — 




Miscellaneoas poipliyries 

No. of 




































The general averages bring out the fact that, while rocks of each group 
may vary considerably among themselves, each group as a whole fits into 
a logical place in relation to the other groups. The established order 
appears to be, most gas from those rocks which contain the greatest pro- 
portion of ferromagnesian minerals. Though much influenced by other 
conditions, such as relative age and nature of the igneous mass, the general 
deduction may be made that the volume of gas obtained from rocks 
varies, in a rough way, in proportion to the percentage of ferromagnesian 
minerals present. Diabases, basalts, and basic schists take first rank in 
the quantity of gas evolved. Next to them appear diorites and gabbros 
which are also near the basic end, but formed under different conditions. 
Andesites are out of their place in this list, as they take precedence over 
granites in the proportion of ferromagnesian minerals, but these andesites 
were all either of Tertiary or Recent age, whereas most of the granites came 
from Pre-Cambrian formations, and, as the next table will show, ancient 
igneous rocks yield more gas than modern ones. The rhyolites, which com- 
bine a scarcity of basic minerals with Tertiary age, foot the list. 

It is to be noted that the rank of a type of rock on the basis of an 
individual gas does not in all cases correspond to its rank for some other 
gas, or in respect to total volumes. The andesites tested gave more carbon 
dioxide than either the granites or the syenites, though both of these types 
greatly surpassed the andesites in the matter of hydrogen. But this in- 
volves another factor: in deep-seated rocks, hydrogen and carbon dioxide 
are of about equal importance; in surface flows, carbon dioxide predomi- 
nates. Though carbon monoxide and methane are somewhat variable, 
the minor gases generally increase or decrease with the total volumes. 

Table 14. — Rocks of sedimentary origin. 


Type of rock. 

No. of 











flhales (non-bitomlnoDS) 

Metamorphosed sediments 

Sandstones and quartsites 













Among sedimentary rocks, sandstones and quartzites yield less gas 
than shales, while the metamorphic group, comprising both altered shales 
and sandstones, together with modified limestones, take an intermediate 
position, though they surpass shales in hydrogen and the sulphur gases. 



Tabls l6,—MdecrU69. 



Ttp* of meteorite. 


Withoat 80| of OigoeU 

Neglecting Anra 

No. of 



























A comparison of the two types of meteorites indicates that carbon 
dioxide is much more important in the gas from stony specimens than in 
that from the metallic bodies, but that iron meteorites yield several times 
as much carbon monoxide and hydrogen as do the stones. Sufficient data 
are not at hand to permit a comparison of the amount of marsh-gas from 
these two t3rpes; nitrogen, however, appears to come in greater volume from 
the iron meteorites. 


Table 16. — Igneous rocks. 







Recent Utsb 

Total Pre<;ambrian 
Grand total 

No. of 

































In addition to those rocks which could be classed either as Archean or 
Proterozoic, there were others which could only be called Pre-Cambrian; 
they are included under the head of Total Pre-Cambrian. 

The rapid and steady decline in the quantity of every gas, in passing 
down the columns from the Archean through the Proterozoic and Tertiary 
to Recent lavas, is very striking. These differences may be due to a com- 
bination of causes. The older rocks may yield more gas than the recent, 
owing to metasomatic changes which have been slowly taking place within 
the rocks. If this be so, the analyses indicate that this process is progress- 
ing at an exceedingly slow rate. Or the early magmas may have been more 
highly charged with gas, some of which has escaped as they were worked 
over and over and brought to the surface in later times. Both of these 
processes have probably been operative. 

Table 17. — Sedimentary and meUi-sedimerUafy rocks. 


Age of rocki. 

No. of 

































1 In this claasification of analyses by the age of the rocks, and in the following one 
based on granularity, only my own analyses have been used. 



Age appears to make less difference in the gas evolved from sedimentary 
or metansedimentary rocks than it does in the case of igneous rocks. All 
of the Proterozoic specimens were of metamorphic types, while only one 
of the Paleozoic sediments had been metamorphosed. The Mesozoic repre- 
sentative was a Jurassic shale altered by an intrusive. The unusual amount 
of sulphur gas in the Proterozoic list is due to two weathered rocks which 
contained iron sulphate. However, even with these omitted, the hydrogen 
sulphide is abnormally high in the rocks of this age. One of the Paleozoic 
shales was so calcareous as to yield 9.28 volumes of carbon dioxide, which 
accounts for the large quantity of this gas. The two bituminous shales 
(analyses 41 and 42) are not included in these averages, since their exces- 
sive volume of gas from organic sources would so influence the figures as 
to disguise some of the characteristics of the other rocks. 


Table 18. — Igneous rocks. 








Varioos porphyries ( mostly Ter- 

No. of 

























From this table it would appear that the fine-grained rocks g^ve off 
more gas than those of coarser granularity. One of the reasons for this 
difference probably lies in the fact that metasomatic changes are favored 
in fine-grained rocks, whose crystals, being smaller, afford more numerous 
junction-planes between the crystals, through which solutions more readily 
traverse the rock than in the coarse-grained varieties. Among other 
changes, hydration and carbonation should alter fine-grained rocks more 
effectively than coarse-grained ones. 

Fineness of grain in igneous rocks usually means that the lava cooled 
rajndly, and this would hinder the escape of the inclosed gas. But in the 
process of slow crystallization, such as produces large crystals and coarse 
texture, much more of the gas would be likely to be crowded out of the 
growing crystals. However, as a general rule, fine-grained igneous rocks 
are siurface flows, while coarse-texture types were formed at some depth 
below the surface, and hence a larger proportion of whatever gas was 
expelled from the rapidly cooling lavas would be more likely to escape 
altogether than would be the case with the gas which was excluded from 
growing crystals in deeper horizons, as in bathylithic intrusions, where final 
escape was difiScult. In this problem of granularity, as in the matter of age, 
the quantities of gas evolved are probably determined by a combination of 
complex factors rather than by any sin^e cause. 


The different gases are not all expelled from rock material at the same 
temperature, nor are they evolved at the same rate. In general, hydro- 
gen sulphide and carbon dioxide are not only the first gases to appear, but 



they are more rapidly g^ven off than the others. Carbon monoxide follows 
the dioxide as the temperature is raised, and generally increases in relative 
importance, as the latter begins to subside, toward the end of the com- 
bustion. Hydrogen and marsh-gas are most conspicuous at high temper- 
atures, and hence attain higher percentages in the last half of the gas than 
in the first portion. Nitrogen appears to be disengaged with much diflScuIty, 
requiring considerable time at an elevated temperature. These general 
facts may be graphically represented by plotting the curves based on the 
experiments with the Baltimore gneiss.^ (See fig. 1.) 







100 200 

300 400 500 




Fio. 1. — ^Plot of ounres representiiig volume of eaeh gas per volume of roek 
obtained at different temperaturee from Baftimore gneiae. 

Nitrogen is omitted from this diagram owing to an imfortunate leak- 
age of air during a part of the experiment, which was sufiScient to vitiate 
the results for this gas. 

^ For tables, see pp. 36-37. 



To determine how much gas might be reabsorbed after being expelled by 
heat, it was thought desirable to use a rock capable of producing a large 
volume of gas. For this purpose a diabase from Nahant, Massachusetts, 
which yielded 13.9 volumes of gas, was selected. This material was heated 
at full blast until the gas evolution had practically ceased, which required 
about four hours; 182 cubic centimeters of gas were obtained. After allow- 
ing the powder to remain in the vacuum overnight, it was removed and 
still more finely pulverized in an agate mortar. It was then submitted to 
forced heat, yielding an additional 20 cubic centimeters of gas in six hours. 
On the third day the powder gave up but 1 cubic centimeter in four 
hours. As practically all the gas available under these conditions was now 
removed, the heat was turned off, and 132.01 cubic centimeters of this gas 
(at 27.0^ and 758 millimeters) immediately introduced into the combustion- 
tube, which was allowed to cool. At the end of 43 hours 101.84 cubic 
centimeters (at 20.0^ and 750 millimeters) remained to be pumped out. 
This being equivalent to 103.73 cubic centimeters at 27.0^ and 758 milli- 
meters, leaves 28.28 cubic centimeters as the volume of gas absorbed by 
the powder. The material in the tube was now heated for 2) hours, but 
only 3.47 cubic centimeters could be extracted before the gas evolution 
ceased. Of this, carbon dioxide constituted more than 85 per cent. There 
still remain 24.81 cubic centimeters lost in the operation — ^a loss which is 
probably to be attributed to the oxidation of that quantity of hydrogen 
to water by ferric oxide, while the tube was cooling. This water-vapor 
being removed by the calcium chloride drying-tube, hydrogen could not 
be again freed by the reverse reaction when the tube was reheated. The 
carbon dioxide may be explained b}"^ carbonation of iron or calcium and 
the subsequent decomposition of these carbonates when heated the second 

In order to ascertain how much absorption there might be at ordinary 
temperatures, 72 cubic centimeters of the remaining gas, from which the 
carbon dioxide had been removed, since carbonation is a recognized proc- 
ess, was allowed to stand in the tubes for eight days. At the end of this 
time no appreciable quantity of the gas had been absorbed. From this 
and the preceding experiment, it is quite evident that while rock material 
may take up certain gases while cooling from a higher temperature under 
special conditions, at ordinary temperatures absorption, if it goes on at 
all, takes place very slowly. Reversible chemical reactions undoubtedly 
play an important part in such absorption as takes place under changing 

Professor Dewar experimented with celestial graphite to ascertain its 
absorbing power for certain gases. After exhausting the graphite of its 
gases, dry carbon dioxide was drawn through the tube for twelve hours 
at ordinary temperatures. The tube was then heated and about 1.1 vol* 
umes of gas, containing 98.4 per cent carbon dioxide, pumped off. The 
graphite on the first heating had given 7.25 volumes of gas, of which 91.8 
per cent was carbon dioxide. Dry marsh-gas was next passed over the 



powder for twelve hours; upon heating, only 0.9 volume, containing 94.1 
per cent carbonic acid, was obtained. The same experiment repeated with 
hydrogen gave only 0.17 volume, in which carbon dioxide reached 95 per 
cent.^ From these figures it would seem that absorption is not very itor 
portant. The steadily decreasing volumes of gas with each successive 
heating show the difiSculty with which the gas is expelled, for apparently 
it is liberated more readily after an interval of time than if reheated imr 
mediately. Hence, unless the material used be completely deprived of its 
gas, there is always a danger in assigning to absorption what may, in reality, 
be only the last portions of the original gas. 

Wright used another method in testing the hypothesis that the gas 
obtained from meteorites has been derived from our atmosphere by a 
process of absorption. He believed that if the gas be due to absorption 
from the earth's atmosphere, a meteorite should have stored up more of 
it after being exposed for a considerable period than shortly after its fall. 
His original analysis of the gas from a meteorite which fell in Iowa Coxmty, 
Iowa, on February 12, 1875, was made a short time after its fall. A year 
later, to extract the same quantity of gas from another fragment of the 
same meteorite required not only a longer time than in the first analysis, 
but more intense heating as well.' if any difference actually existed, a 
loss rather than a gain was indicated in this interval. 

To test the effect of air exposure on a rock powder which had previously 

been heated until the gas evolution had completely ceased, the exhausted 

powders of my investigation were kept stored in paper bags, and several 

of them were reheated after intervals of some months. Two analjrses of 

the iron basalt from Ovifak, Greenland, made 10 months apart, were as 


Table 19. 



No. 45. 

Analysis after 
10 months. 

Hydrogen solphide 

Carbon dioxide 




8.05 vols. 






1.76 vols. 

Carbon monoxide 





This basalt yielded about one-third as much carbon dioxide after the 
interval as it did when originally heated, but the hydrogen in the second 
portion of gas was almost a negligible quantity. 

A second test was made mth a chloritoid schist from the Black Hills, 
after an interval of more than a year. 

> Sir James Dewar, Proc. Roy. Inst., vol. 11 (1886), p. 647. 
* Wright, A. W., Am. Jour. Sd., vol. 11 (1876), p. 262. 



Table 20. 

No. 28. 

a year later. 

Hydrogen sulphide 

Our bon dioxiae 






3.72 vols. 





• • • • 

2.69 vols. 

Carbon monoxide 

Methane r . . . 


■■-*j7 ^■**'^^**"* ••••••••4,. »..•••• 



In this case, hydrogen has been restored somewhat more completely 
than carbon dioxide. Both of them amount to approximately two-thirds 
of the original volume of these gases. 

A third test was made with amphibolite, after an interval of four months. 

Table 21. 

No. 94. 

Analysis four 
months latex. 

Hydrogen salphide .' . . . 

Carbon dioxiae 




6.40 vols. 






3.05 vols. 

Carbon monoxide 





The recovery is here more marked in the case of hydrogen than in that 
of carbon dioxide. 

A foarth test was made with Keweenawan diabase from Houghton, 
Michigan, after an interval of six months. 

Table 22. 

No. 85. 

Analysis 6 
months later. 

Hydrogen sulphide 

Carbon dioxiae 





3.88 vols. 






1.92 vols. 

Carbon monoxide 





After reposing six months in a paper bag, this diabase gave as much 
carbon dioxide, when heated, as it had in the first combustion; but less 
than one-fifth as much hydrogen was evolved on the second heating. 

It is clear that an interval of time partially restores the gas-producing 

properties of these rock powders. For this phenomenon, there are two 

possible explanations. Either the first heating does not expel all of the 

gas contained in the rock, which, by some sort of diffusion or molecular 




rearrangement, gradually prepares itself to come off when again heated, or 
else the rock powder absorbs gases from the atmosphere. If the carbon 
dioxide were derived from the decomposition, at high temperatures, of a 
carbonate such as that of calcium, the oxide of calcium thus produced 
would be likely to capture carbon dioxide from the air, though perhaps 
this would be a slow process in a paper bag where the circulation of w' was 
comparatively limited. Also, if the hydrogen came from chemical reactions 
between ferrous salts and water combined in hydrated minerals, the atmo- 
sphere might have restored to these minerals some of the water which they 
lost when first heated. It was thought that rehydration, if combined 
water be a vital factor in the production of hydrogen, could be more readily 
effected by placing the exhausted powder in water for a few days than by 
wrapping it up in a paper bag for as many months. 

Accordingly, the Keweenawan diabase powder (No. 85) which ori^n- 
ally gave 3.88 volumes, and after six months 1.92 volumes, was heated a 
third time (a week later) with the evolution of very little gas. This powder, 
after cooling in the vacuum, was taken out of the combustion-tube and 
immediately placed in a flask filled with freshly distilled water. A stopper 
being fitted into the flask, it was allowed to stand for 66 hours. At the 
end of this time, the water was poured off, the powder quickly, but thor- 
oughly, dried and put into the combustion-tube. When heated, this powder 
gave off 0.79 volume of gas; but instead of being largely hydrogen, 67.72 
per cent of this was carbon dioxide. Hydrogen amounted to only 14.69 per 
cent, while carbon monoxide reached 15.06 per cent. An analysis of this 

gas gave: 

^ ^ Tabi^ 23. 



Carbon dioxide 









Carbon monoxide 





This carbon dioxide could not have come from the air, but must have 
existed within the material and must have withstood three successive 
heatings in the combustion-tube. From a comparison of these figures 
with the two previous analyses of the gas from this material, what is true 
of the carbon dioxide would appear to be true of the hydrogen as well. 
This experiment favors the conclusion, that the gas which is obtained from 
a rock powder by a second heating after a period of time, is not due to a 
process of selective absorption from the atmosphere, but rather to changes 
which have been slowly taking place within the powder itself. 

However, the results of these experiments upon the absorption of gas 
by rock powders at ordinary temperatures and pressures can not throw 
much light upon the source of the gases, or how they came to be embodied 
in the rocks, since the conditions under which the rocks were formed must 
have been very different. While high temperatures, in general, tend to 


expel the gaseous constituents of the rocks, high pressures would have the 
effect of promoting absorption. Moreover, it is possible that molten 
lavas might absorb, or dissolve, certain gases without an increase of pres- 
sure. But the testimony of volcanic gases and of the scoriaceous surfaces 
of lava flows favors the idea that gases and vapors are constantly being 
boiled out of molten lavas whenever exposed under the ordinary atmo- 
spheric pressure. Lavas give off gas rather than absorb it, at the earth's 
surface; however, at considerable depths below the surface the action may 
be entirely different. If the conception be entertained that the earth's 
interior is, for the most part, solid with only threads of liquid lava here 
and there, the question for this solid portion would be one of the ability 
of great pressure to cause a solid to absorb gases. This need not be further 
dwelt upon, since most of the igneous rocks which are accessible have been 
in the liquid state at some time. In the case of the threads of liquid magma 
there is reason to suppose that gas, if it could be brought into contact 
with this lava, would become incorporated in it owing to the great pres- 
sure. But this does not explain the original source of the gases, nor how 
they can be brought in contact with the liquid rock under the prevailing 
conditions of temperature and pressure. 


In order to explain the immediate source of the gases obtained by 
heating rock material in vacuo, three different hypotheses naturally pre- 
sent themselves. The simplest of these is to suppose the gases to exist in 
minute cavities or pores, having been entrapped within the rock during the 
process of solidification. This supposition is suggested and supported by 
the observation that microscopic slides of some minerals, notably quartz 
and topaz, reveal numerous small gas-bubbles. But while there is evidence 
that some gas is thus held in cavities, there is equally strong evidence to 
show that the greater part of it can not be attributed to this source. 

To escape the diflSculties encountered by the first hypothesis, appeal 
is made to the imperfectly understood property of some of the elements 
to "occlude," or dissolve within their mass, certain gases. It is remem- 
bered that under the proper conditions palladium will occlude 900 times 
its own volume of hydrogen, and that the same gas is also absorbed, in 
lesser degree, by other metals, particularly platinum and iron, while silver 
has a similar affinity for oxygen. This principle applied to igneous rocks 
as a hypothetical source of their gases becomes at once a more difficult 
proposition to prove or disprove. 

The third hypothesis, more conservative than either of the others, 
assumes that these gases do not exist in the rocks in the uncombined, or 
gaseous state, but are produced in the combustion-tube by chemical re- 
actions at high temperature. The oxides of carbon and sulphur are assigned 
to the decomposition of carbonates and sulphates; methane to organic 
matter present, carbides, or to high temperature reactions between hydro- 
gen and the carbon gases; sulphureted hydrogen to sulphides; nitrogen 
to nitrides; while hydrogen is liberated from steam by the action of metallic 
iron or ferrous salt. 




The studies of Brewster, Davy, Sorby, Hartley, and others, have 
established the presence of gas, generally carbon dioxide, though sometimes 
nitrogen, in the minute cavities of certain crystals. This has been widely 
known to geologists, and hence, when it was discovered that many crystal- 
line rocks 3deld gas upon heating in vacuo, it was natural to suppose that 
the gas came from cavities. Such was the view taken by Tilden.* But 
while microscopical investigations indicated that carbon dioxide consti- 
tutes more than 90 per cent of the gaseous matter inclosed in these cavi- 
ties, and hydrogen is not found in more than traces, the latter gas is the 
most important constituent of the mixture derived from rocks by heat. 
In addition to this, the observation that those rocks which are not known 
to contain many gas cavities produced several times as much gas as the 
cavernous quartzes also suggested that the bulk of the gas, at least, could 
not be attributed to inclosure in cavities. Moreover, basic rocks were 
found to be more productive than acidic, whereas it had generally been 
supposed that the latter, owing to their greater viscosity, should entrap 
more gas and vapor than the more fluid basic lavas. 

The suspicion that the gas did not come from cavities in any large 
degree was strengthened by the observation that the composition of the 
gas varied according to the temperature to which the rock powder was 
heated. If the gas comes from cavities, its liberation should commence 
with a slight rise of temperature and should continue more or less steadily, 
as the heat increases, until the expansive force of the gas opens up most 
of the pores. Since all gases expand equally, one should burst its con- 
fines as soon as another, and a sample of gas obtained at any given 
temperature should not differ very widely in composition from that evolved 
at any other. 

Neglecting hydrogen sulphide and nitrogen, the character of the gas 
obtained at various temperatures from Baltimore gneiss ' is shown by the 
following table: 

Tablb 24. 




At 540°. 




Carbon dioxide 

Carbon monoxide 


























Or, combining the separate analyses so that each figure represents the 
percentage of the total gas obtained up to the specified temperature, the 
result is as shown in table 25: 

» Tilden, Chem. News, vol. 76 (1897), p. 169; Proc. Roy. Soc., vol. 64 (1897), p. 463. 
' Material of analysis, No. 28. 



TlBLE 25. 


Carbon dioxide . . 
Oarbon monoxide 




Volamee .. 

200(0 360° 












20° to 540° 























Carbon dioxide thus appeared first, constituting 93 per cent of the gas 
evolved at 360® C, while hydrogen was not present in a measurable quan- 
tity. On the other hand, at the highest temperature used (850®) hydrogen 
amounted to 95 per cent of the total and carbon dioxide was entirely 
wanting. The steady decrease in the proportion of carbon dioxide with 
the elevation of the temperature, and the proportionate increase in the 
value of the hydrogen, are striking. The minor constituents, carbon 
monoxide and methane, underwent some variations, but did not change 
so radically. The former came off at the lower temperature, but declined at 
full red heat. I>Htrogen, it appears from other experiments, does not appear 
in the gases obtained by moderate heating, but increases steadily in impor- 
tance when the heat is carried higher. It is the last gas to be liberated. 

The complete table, expressing the volumes of each gas per unit volume 

of gneiss, follows: 

^ Table 26. 








100®. boiliniF water 















218®, boiling naphthalene 

360®, boiling anthracene 

448®, boiline solphar 

540®, metal oath 

600®, dnll red heat 

800®. foil blast 

850®, forced heat 








These results are graphically represented in the curves of figure 1. 

The fact that little gas could be obtained below 450® is in itself a strong 
argument against the hypothesis that the gases come from pores, and 
there also seems no way in which the behavior of the gases, as set forth by 
these curves, can be consistently fitted into that theory. 


Meteorites have alread}' been subjected to investigation of this sort, 
though not with this purpose in mind. Mallet divided the gas which he 
extracted from the meteoric iron of Augusta Coxmty, Virginia, into three 
portions;* his results have been reduced by Wright* to the figures given 
in table 27: 

» Mallet, Proc. Roy. Soc., vol. 20, p. 367. * Wright, Am. Jour. Sci., vol. 2, p. 261. 



Table 27. 





Beginning . , , , - t 









The analyses of meteorites by Wright show that, in all cases, carbon 
dioxide reached a higher percentage in the gas evolved at 500® than it 
did in that obtained at red heat, and that the reverse of this was true of 
hydrogen in the stony meteorites. In the iron meteorites, however, two 
analyses indicated a marked fall in hydrogen with the increase of heat, while 
the other two were characterized by an increase. Wright's figures for the 
meteorite from Guernsey County, Ohio, illustrate the continuous decrease 
in the percentage of carbon dioxide: At 100®, 95.92 p. ct.; at 250®, 86.36 p. 
ct. ; at 500®, 82.28 p. ct. ; incipient red heat, 33.55 p. ct. ; red heat, 19.16 p. ct. 

The volume of gas obtained at each temperature is only stated for 
500® and red heat. These show that up to 500®, 2.06 volumes were evolved, 
and that above this point only 0.93 volume was received. From this it 
appears that the diminishing percentages of carbon dioxide above 500® 
represent an absolute slackening of the output of that gas, as well as an 
apparent decrease due to the greater evolution of hydrogen. It might 
be argued that in this case, where gas was produced at only 100®, the 
cavities contributed the carbon dioxide, yielding it early and then slacken- 
ing, as would be expected; but even if this be admitted, the hydrogen 
manifestly can not be ascribed to that source. Tending in a measure to 
support this view is the work of Sorby,* who has shown that olivine crystals 
in the meteorites of Aussun and Parnallee, when examined under the micro- 
scope, contain numerous small cavities filled with gas, similar to those which 
have been observed in many terrestrial minerals. 

In his earlier paper, Wright expressed the opinion that the gases were 
partly condensed upon the particles of iron and partly absorbed within 
them. Later he took the position that while some gas may be condensed 
upon the fine particles of iron, a large part of the carbon dioxide, and prob- 
ably also of the other gases, is mechanically imprisoned in the substance of 
the meteorite. This view, which does not seem to be in accord with his 
researches at different temperatures, he bases largely upon a single experi- 
ment. Material from the Iowa meteorite was finely pulverized and the 
iron grains separated from the non-metallic powder. A third portion 
consisted of coarse fragments of the meteorite. The three portions heated 
for the same time gave the following results: 

Tablb 28. 









0.97 1..g 

0.61 / ^-^^ 




» Sorby, Proc. Roy. Soc., vol. 13 (1864), pp. 333-334. 



The greater volume of gas from the fragments was taken to indicate 
that a portion of the gas was lost in the process of pulverization. An 
analysb of these figures reveals the fact that the difference in volume 
was chiefly due to the deficiency of the combined portions in hydrogen, 
instead of carbon dioxide, and that while there was also a slight loss of 
the latter gas, there was a decided gain in nitrogen. 

Returning to the rocks, Tilden ^ is authority for the 
statement that it does not make much difference in the 
quantity of gas evolved, whether the material be taken 
in chunks or in a fine powder. Instead of abandoning 
the idea of cavities, he believed them to be very minute. 
But this is approaching an alternative hypothesis; if 
the reduction of the cavities is carried far enough — to 
intermolecular spaces — practical occlusion is the result. 

Another objection to the theory of mechanically-re- 
tained gases apparently exists in the slowness with which 
the gas is liberated when the material is heated. Usually 
about three hours and often a very much longer time is 
required to expel the gas. Unless the gas from cavities 
be assumed to escape by diffusion through the walls of 
the inclosing mineral, instead of violently bursting its 
confines, there is no reason why it should not come off 
with a rush when the combustion-tube is heated rapidly 
to redness. Some rocks, generally those yielding a mod- 
erate quantity of gas in which carbon dioxide is the 
principal constituent, often give up their gas quickly — 
mostly within the first 60 to 90 minutes, although the 
generation continues for a longer time, before ceasing 
altogether. But other varieties of rock, particularly 
those noted for greater volumes, in which the percent- 
age of hydrogen runs high, emit gas slowly and steadily 
for three or four hours. 

These considerations led me to try a series of experi- 
ments which should show how much gas actually could 
be obtained from the opening of cavities alone. For this 
purpose a crusher was devised (fig. 2), capable of pulver- 
izing a rock specimen in a complete vacuum. Adopt- Fio.2.-Appar»tu8for 
ing the principle of the familiar steel mortar, this was ^^j^rro&BpttimmB 
constructed in three pieces. The cylindrical cup in 
which the rock material is crushed possesses an internal diameter of 7 
centimeters and a depth of 9 centimeters. The walls are purposely 
made thick and strong and the bottom is protected from the abrasion of 
hard minerals by inserting a disk of hardened steel. Inserted in the walls 
is a stopcock through which the apparatus is to be exhausted and the 
gases later pumped out. A circular steel cap, or cover, provided with six 
screws, whose sockets are depressed in the top of the cylinder, is intended 
to make the chamber of the mortar air-tight. In the center of the cap- 

» Tilden, Chem. NewB, vol. 75 (1897), p. 169; Proc. Roy. Soc., vol. 64 (1897), p. 453. 


piece is a hole large enough to permit the ready movement of the piston- 
shaft. Around this hole on the upper side there is welded a short piece of 
steel tubing which is to guide the piston-rod and serve as a place of attach- 
ment for the rubber tubing in which the shaft of the piston is incased. 
The piston is a shaft 50 centimeters in length, 2.2 centimeters in thickness^ 
to which is attached a head piece of hardened steel which will fit snu^y 
into the cylinder. Near the upper end of the piston is a cross-bar serving 
as a handle, and also a flange to which the rubber tube is to be fitted. 

When ready to put together, the piston-shaft is incased in a 14nch 
tube of pure rubber, 45 centimeters long, which is tightly fitted and wired 
to the flange near the end of the rod, whereupon the other end of the shaft 
is slipped through the hole in the cover-piece, and the piston-head affixed. 
The lower end of the rubber tube is wired to the steel tube of the cover- 
piece which, after the rock specimen has been placed in the cylinder, is 
fitted with a rubber washer and screwed as tightly as possible to the cylinder. 
The rubber tube is taken of length sufficient to allow the head of the piston 
to touch the bottom of the cylinder; by pulling upward on the handle the 
rubber wrinkles and folds upon itself, affording ample play to the piston. 

The stopcock is connected with the mercury-pump and the cylinder of 
the crusher exhausted, after which vigorous strokes delivered at the end of 
the pbton with a heavy mallet crush the rock, thus opening the gas cavities. 
Whatever gas is liberated, is pumped into the receiver and analyzed in the 
ordinary way. 


Of the first rock tested, a basalt from the Faroe Islands, 42 grams 
were crushed finely enough to pass through a 30-mesh sieve, besides several 
times as much, less completely pulverized. In all, less than 0.1 cubic 
centimeter of gas was obtained, which may be considered as practically no 
gas at all, since this small quantity is within the leaking possibilities of the 

A slightly scoriaceous basalt from Hawaii produced about 0.1 cubic 
centimeter of gas, which appeared to be largely air. No carbon dioxide 
could be detected. Of this basalt, 18.3 grams passed through the sieve. 

15.73 grams of vein quartz from Utah (No. 71 of the analyses) gave no 
trace of gas. 

In an effort to demonstrate conclusively that the lack of gas liberated 
by crushing these lavas was not due to defective apparatus, a glass bulb 
of measured capacity, filled with air, was broken in the crusher in place of 
the rock ordinarily used. The result showed that gas introduced into the 
crusher can be extracted without sensible change in volume. As diffusion 
through the rubber tubing was considered a possible, though not very 
probable, source of error, a further trial was made, using hydrogen, lightest 
and most active among the gases, in order to put the apparatus to as 
severe a test as possible. The purity of this hydrogen had previously been 
established by analysis. The bulb broken, the gas was pumped off and 
exploded with air. The observed shrinkage agreed, within the limit of 
error, with the amount of hydrogen calculated to have been contained 
within the bulb. 


Being desirous of finding some specimen which would yield gas when 
crushed in this manner, I procured some crystals of cavernous quartz 
from Porretta, Italy, in which several of the cavities exceeded a millimeter 
in diameter. 5.91 grams were crushed to sufficient fineness to pass through 
the sieve, and 61.66 grams were partially crushed. 0.08 cubic centimeter 
of carbon dioxide was obtained, which, supposing that it all came from the 
5.91 grams, would be equivalent to only 0.03 of the volume of the quartz. 
An analysis showed also a little methane and some nitrogen, but the amount 
of gas available was too small for the determination to be of any value. 

The result of this last test agrees with the microscopic studies of the 
early investigators. Carbon dioxide exists in the cavities of quartz, but 
its volume, compared with the volume of the inclosing mineral, is small. 
Microscopical observations seem to show that gas cavities occur almost 
exclusively in a certain set of minerals which combine hardness usually with 
imperfect cleavage, namely, quartz, topaz, garnet, spinel, beryl, chrysoberyl, 
corundum in the form of rubies, sapphires, and emeralds, and diamond. 
These are minerals which, once they had inclosed gas, would hold it, even 
under great pressure. 



The double series of iron salts, ferrous and ferric, together with the 
intermediate ferroso-ferric compounds, reacting with oxidizing or reducing 
agents, undergo various reversible reactions whose possibilities are great. 
When steam is passed over metallic iron or ferrous oxide at a red heat, it 
is decomposed, giving up oxygen to the iron, and at the same time pro- 
ducing free hydrogen. The reactions may be written: 

Fe + HjO = FeO + H, 3FeO + HjO = Fefi^ + H, 

Hydrogen is produced in this way most rapidly at temperatures about 
500®. Stromeyer is authority for the statement that the breaking up of 
water begins at 150° but takes place very slowly; at 200° somewhat more 
rapidly; at 360° the process requires several hours; at 860° it is complete 
in less than one hour; while near the melting-point of iron several minutes 
are sufficient.^ 

The authorities agree that ferric oxide is not formed in this process; 
the magnetic oxide, FegO^ is the final product of the action of a current 
of steam upon ferrous oxide.^ 

But these reactions are completely reversible. According to Gay- 
Lussac, magnetite is reduced to the metal by hydrogen at every tempera- 
ture between 400° and the highest degree of heat obtainable in the com- 
bustion-furnace, particularly at the same temperature at which steam is 
split up by glowing iron.' Siewert states that ferric oxide (from the oxa- 
late) is not altered by hydrogen at 270° to 280°; between 280° and 300° 

* Stromeyer, Pogg. Ann., vol. 9, p. 476. 

'Among others, Kegnault, Ann. de Chim. et Phys., vol. 62, p. 348. 

* Gay-Lussac, Ann. de Chim. et Phys., vol. 1, p. 33. 


it is reduced to ferrous oxide, and when heated above 300^, to the metal.^ 
The more recent studies of Moissan give different figures ;* ^^,0, is reduced 
by hydrogen at 300° to Fe^O^ in 30 minutes; at 500"* to FeO in 20 minutes; 
at 600° to 700° to metallic iron. 

If the hydrogen or water-vapor produced by these reactions is not 
removed, the process continues only until a condition of equilibrium is 
established. In extracting the gases from rocks, the products of these 
reactions were rapidly removed, so that final equilibrium was probably 
never attained. Hence, in these experiments the direction in which thd 
reaction will proceed depends upon whether there is ferrous oxide and 
water, ox ferric oxide and hydrogen, most abundantly stored in the rock. 
Ferrous and ferric salts behave, in general, like the oxides. 

Since most igneous rocks contain ferrous as well as ferric salts, the 
possibility that, when heated in the presence of steam, hydrogen will be 
produced, must always be taken into account. In terrestrial rocks water 
of constitution is generally present and often is not expelled below a bright 
red heat. Thus, a rock containing a ferrous compound in appreciable 
amount, together with water of crystallization, a portion of which is re- 
tained up to red heat, will be in a condition to furnish hydrogen upon the 
application of heat. 

In general, the analyses show that the greater the amount of iron present 
in the rock, the more hydrogen may be expected. This may be the result 
of chemical action, or a selective occlusion of hydrogen manifested by iron 
and its compounds. Magnetite, being the end product of the reaction of 
water upon iron, can not produce hydrogen by this chemical interaction, 
though it might possess the occlusive properties of iron compouinds. Anal- 
ysis of the black sand from the bed of the Snake River, Idaho,' indicates 
that iron in the form of magnetite does not yield much hydrogen. How- 
ever, these figures have no great significance, for, even though an abundance 
of hydrogen existed in the ore, either occluded or mechanically imprisoned, 
the magnetite would, at red heat, quickly oxidize it to water, with the 
exception of a small portion of free hydrogen maintained by the reverse 
reaction. The analyses show that basic diabases and basalts yield the 
most gas, while acidic rhyolites give but little. These are also among the 
maximum and minimum iron-bearing lavas. But the difference in hydro- 
gen is much greater proportionately than the difference in ferrous salts. 
Table 13 ^ also shows that andesites, which are nearer the basic end of the 
scale than the acidic, do not greatly exceed the rhyolites in hydrogen. 
The difference between the two types of rocks, acidic and basic, in point 
of volume of the individual gases, while somewhat more conspicuous in 
the case of hydrogen, is generally true of the other gases as well. 

Endeavoring to prove that the hydrogen obtained by heating minerals 
came entirely from chemical reactions, Travers experimented with the 
secondary mineral chlorite, calculating how much ferrous iron should have 
been oxidized to give the quantity of hydrogen and carbon monoxide 
evolved.* This he found to agree closely with the difference in amount of 

* Siewert, Jahresbericht d. Chem., 1864, p. 266. * Ante, p. 27. 

* Moissan, Comptes Rendus, vol. 84, p. 1296. » Travers, Proc. Roy. Soc., vol. 64, p. 132. 
' Analysis No. 46. 



ferrous iron present before and after heating. Another test with feldspar 
from the Peterhead granite not showing correspondence, seemed to Travers 
to be explained by the presence of both metallic iron and ferrous oxide in 
the feldspar. The presence of a considerable amount of metallic iron in 
a feldspar which crystallized from an acidic magma containing an excess 
of ffllica is quite unusual. This feldspar treated with dilute sulphuric 
acid yielded about four volumes of hydrogen. 

Against the theory that the hydrogen was largely derived from the 
action of water-vapor on ferrous compounds, may be placed the very 
marked change in color which the rock undergoes during the process of 
heating. I have observed that whenever a rock powder, before being 
placed in the tube, possesses an orange, brownish, or reddish tint due to 
ferric oxide, the combustion invariably alters the tone to a greenish gray. 
This suggests a reduction of ferric oxide to ferrous oxide, a process con- 
suming hydrogen. In order to test this question, a specimen of bright-red 
Permian sandstone from the Garden of the Gods near Colorado Springs^ 
was powdered. These Red Beds are supposed to consist of thoroughly 
oxidized material; this opinion was partially confirmed by chemical tests 
which gave a weak reaction for ferrous iron, but indicated much ferric. 
After heating, the brick-red sand had become dull gray-green in color. The 
gray sand from the combustion-tube gave a stronger reaction for fer- 
rous iron. Later, a quantitative determination of the ferrous iron present 
before and after heating was undertaken. Equal weights of the two sands 
were boiled with strong sulphuric acid ' for two hours and then allowed to 
stand overnight. In each case the solution was effected in an atmosphere 
of carbon dioxide to prevent oxidation by oxygen from the air. The two 
solutions were then titrated with potassium permanganate solution. 3.09 
grams red sand required 1.97 cubic centimeters N/10 EMn04; 3.09 grams 
gray sand required 2.52 cubic centimeters N/10 KMn04. ^-^^ cubic centi- 
meter N/10 EMn04 is equivalent to 0.015 gram of iron, which is the 
weight of the metal reduced from the ferric to the ferrous state. For the 
total weight of sand used in the gas analysis (85 grams), the increase in 
ferrous iron should be 0.423 gram, which would correspond to an oxidation 
of approximately 85 cubic centimeters of hydrogen. Yet both hydrogen 
and carbon monoxide were obtained from this sandstone in considerable 

Table 29. 



Hydrogen sulphide 

Gurbon dioxide 








Carbon monoxide 





' Analysis No. 78. 

' 3 parts cone, acid to 1 part water. 



In order to ascertain the quantitative effect of the presence of ferric 
oxide in moderate amount, 0.77 gram of pure Fe,0, was mixed with 20.04 
grams of diabase powder, tinting thb latter a reddish brown. An analysis 
of the resulting gas and of the original diabase gave the figures shown in 
the following table: 

Table 30. 

Retalting gaa. 

Original <UAbMe.i 



Per cent. 


Hydrogen sulphide 

Carbon dioxiae 












Hydrogen sulphide 

Carbon dioxide 











Carbon monoxide 


Carbon monoxide 












1 Analysia No. 86. 

A comparison of these results shows that, while the yield of hydrogen 
was diminished by the ferric oxide to less than half of what it would have 
been, the carbon monoxide was not affected. The ferric oxide apparently 
only went down to a state of equilibrium, and was not in suj£cient quantity 
to offset the copious evolution of hydrogen from the diabase. The brown 
color, however, was replaced by green. 

To get rid of the iron, and particularly ferrous iron, material from the 
same diabase specimen was treated with concentrated nitric acid for 66 
hours. Much gas came off at first, nitric oxide, perhaps from the action 
of the acid on pjrrite, being very conspicuous. The powder, washed repeat- 
edly on a filter until all the acid had been removed, was dried in an oven 
overnight and then heated at 115^ in an air-bath for half an hour. Two 
and a half hours at red heat, in vacuo, then expelled only 0.23 volume of 
gas from the diabase powder. Its composition is given in table 31. 

Table 81. 



Hydrogen sulphide 

Carbon dioxide 






Carbon monoxide 





A similar test was made with dilute sulphuric acid, in a vacuum. In 
this experiment, the gas driven off by the acid during the first 2 J hours 
was collected and analyzed. Table 32 shows this to have been chiefly 
carbon dioxide. 



Table 32. 



Hydrogen sulphide 

Carbon dioxide 









Carbon monoxide ) 

Methane / 




As a precautionary measure, to avoid the introduction of any metallic 
iron in the process of pulverization, the diabase was reduced to a powder 
in an agate mortar. The brass sieve was not used. Hence this hydrogen 
did not come from any action of the acid upon a metal introduced during 
the manipulations. 

This powder, after remaining in a vacuum with an excess of sulphuric 
acid for three days, was washed thoroughly on a Gooch filter until the last 
traces of calcium sulphate had been removed. After drying for an hour 
at 125^, the powder was placed in the combustion-tube and heated to 
redness. The sulphuric acid left more gas in the rock than the nitric. 

Tablb 33. 



Hydrogen Bolphide 

Cflffoon dioxide ^ . x ^ ....... a 










Carbon monoxide 

Methane ............ t t t r x r 




From these experiments it would appear that acids remove the critical 
gas-producing factors without liberating a notable amount of any gas 
except carbon dioxide. Whether hydrogen may not pass into solution 
with the iron, without being freed, is a question which naturally arises, 
but the balance of chemical opinion is against this supposition. 

Professor Dewar digested celestial graphite in strong nitric acid for 
several hours and, after washing and drying, found that with heat it gave 
exactly the same amouint of hydrogen as before treating with the acid. 
This would suggest that, in the case of celestial graphite, the hydrogen 
was not connected with iron, but existed in some very stable form.^ 

If all the hydrogen was produced by the reaction of water on ferrous 
salts, it would seem as if the volume obtained should bear a direct relation 
to the quantity of these two critical constituents present in the rock. 
To throw light on this matter, two rocks of the same origin, but of different 
chemical composition, presented the most favorable line of attack. An 
intrusive andesite and a specimen of vein quartz derived from the mag- 

^ Dewar, Proc. Roy. Inst., vol. 11, p. 560. 



matic waters of the intrusion, kindly furnished by Dr. G. K. Leith, were 
used to illustrate this point.^ Though containing very different quantities 
of ferrous compounds, they 3delded identical volumes of methane, hydrogen, 
and nitrogen. These analyses are given in table 34. 

Table 34. 





Per cent. 


Hydrogen sulphide 

Gfixbon dioxiae 







Hydrogen sulphide 

Cw*bon dioxide 





Carbon monoxide 


Carbon monoxide 












The great excess of carbon dioxide in the andesite is assigned to car- 
bonation of that lava subsequent to its formation — a process to which the 
quartz would not be susceptible. 

The observation that comparatively pure quartz yielded half a volume 
of hydrogen suggested a quantitative analysis to determine the amount 
of iron actually contained in hydrogen-producing quartz. For the purpose 
quartz from Orange, New South Wales, was selected. 102.72 grams of the 
quartz yielded 4.81 cubic centimeters of hydrogen at 0° and 760 millimeters.' 
After the gas had been extracted, two different portions of the exhausted 
mineral were digested with aqua regia — one of them boiled for an hour, 
the other being allowed to stand during several days, and occasionally 
warmed to the boiling-point. The acid may be considered to have dis- 
solved all the iron from which gas could have escaped. To make the case 
certain, all of the iron detected has been supposed to have existed in the 
quartz as ferrous oxide, although some of it undoubtedly occurred in the 
form of ferric compounds. The iron was weighed as FojOj. 

First determination: 

22.22 gms. quartz contained 0.0015 gm. Fe,Oa 

102.72 gms. quartz would contain 00693 gm. Fe,Oa 

102.72 gms. quartz would contain 00485 gm. Fe 

Fe (as FeO) required to give 1 c.c. hydrogen 00748 gm. 

Maximum amount hvdrogen from reaction 65 c.c. 

Hydrogen actually obtained (at 0^ and 760 mm.) 4.81 c.c. 

Hydrogen not from this reaction 4.16 c.c. 

Second determination: 

52.02 gms. quartz contained 0.0042 gm. Fe,0, 

102.72 gms. quartz would contain 00829 gm. Fe,0, 

102.72 gms. quartz would contain 00580 gm. Fe 

Fe (as FeO) required to give 1 c.c. hydrogen 00748 gm. 

Maximum amount hydrogen from reaction 77 c.c. 

Amount of hydrogen actually obtained 4.81 c.c. 

Hydrogen not from this reaction 4.04 c.c. 

*■ Analyses Nos. 70 and 71. 
* Analysis No. 100. 


According to these two determinations, this quartz evolved respectively 
7.4 or 6.2 times as much hydrogen as could have been generated by the 
reaction 3FeO+HjO = Fe^O^ + H,. 

If the iron existed as pyrite, four times as much hydrogen as could 
come from ferrous oxide might have been produced in accordance with 
the equation 

3FeS,+4H,0 = Fe30, + 4Hj+6S 

On the basis of this equation the excess of hydrogen from the quartz 
is much reduced. 

Fint determination: 

102.72 gma, quartz contain 0.00485 gm. Fe 

Fe Tas FeS,} required to give 1 c.c. hydrogen 00187 gm. 

Hyorogen poesible from reaction 2.60 c.c. 

Hydrogen actually obtained 4.81 c.c. 

Hydrogen not from this reaction 2.21 c.c. 

Second determination: 

102.72 toDB. quartz contain 0.00580 gm. Fe 

Fe Tas FeS,) required to give 1 c.c. hydrogen 00187 gm. 

Hyorogen possible from reaction 3.08 c.c. 

Hydrogen actually obtained 4.81 c.c. 

Hydrogen not from this reaction 1.73 c.c. 

These computations assume not only that all the iron in the quartz 
was combined as pjrrite; and that it was completely oxidized to magnetite, 
but that the hydrogen sulphide produced was entirely dissociated into 
hydrogen and sidphur. But the iodine titration in the gas analysis revealed 
0.36 cubic centimeter (at 0^ and 760 millimeters) of sulphur gas whose 
odor was that of hydrogen sulphide rather than sidphur dioxide. If this 
were H^S, it would diminish the amount of hydrogen which could have come 
from the reaction by 0.36 cubic centimeter; if, however, it were sulphur 
dioxide the volume of possible hydrogen would be swelled by 0.72 cubic 
centimeter in accordance with the reaction 

S+2H,0 = S0,+2H, 

But as there was certainly much more hydrogen sulphide than sulphur 
dioxide absorbed by the potassium hydroxide solution, it will be safe to 
balance the possible SO, formed, by the £[,8 undissociated, and ignore 
these corrections, which would probably reduce, rather than increase, the 
quantity of hydrogen which might result from pjrrite. 

If the iron had all been locked up in the mineral chalcopyrite (CuFeS,) 
the hydrogen might be accounted for, but chemical tests failed to detect 
the copper which this supposition would require. Just how much hydrogen 
might be expected from iron nitride (Fe,N) is not certain, since, in the 
presence of superheated steam, the nitrogen is more likely to unite with 
hydrogen and come off as ammonia rather than as free nitrogen, and 
ammonia is not dissociated short of the electric spark. That most of the 
iron in the quartz is in the form of a nitride is highly improbable. Iron 
carbide also woidd not 3deld sufficient hydrogen. 

Another mineral apparently containing very little iron, but which 
3delded considerable hydrogen, was the beryl of analyses 101 and 101a. 


Though as transparent as window-glass, one volume of this beryl con- 
tributed 0.31 volume of hydrogen. A determination of its accessible iron 
was made by pursuing the same method as was used for the quarts. The 
results were: 

35.00 gms. beryl contained 0.0003 gm. Fe^, 

127.52 gms. beryl would contain 00109 gm. Fe,0, 

127.52 sms. beryl would contain 00076 gm. Fe 

Fe (as FeO) required to give 1 c.c. hydrogen 00748 gm. 

Maximum amount hydrogen from reaction 0.10 c.c. 

Hydrogen actually obtained (0^ and 760 mm.) 14.S9 c.c. 

Hydrogen not from this reaction 14.79 c.c. 

This beryl expelled nearly 150 times as much hydrogen as can be 
assigned to the interaction of steam and ferrous oxide under the most 
generous assumptions. The actual hydrogen is 37 times the maximum 
quantity possible from this weight of iron, either as pjrrite or in the metallic 
state. Here is a very declared case demonstrating the inadequacy of chem- 
ical reactions involving iron to generate the hydrogen obtained. 

Heated in a closed tube with a limited amount of air, beryl is known 
to give up a small quantity of water which, in some varieties of the mineral, 
may reach 2 per cent. The question whether the excess of hydrogen over 
that possible from reactions between water and iron coidd have arisen 
from the dissociation of this water is easily answered. The recent researches 
of Nemst upon the dissociation of steam indicate that, at temperatuies 
below 2000^ C, the process takes place only to a very limited extent. 
At 1124^ C, which is somewhat above the point to which the beryl was 
heated, only 0.0078 per cent of the total steam can be dissociated.^ At 
this temperature, 127 grams of beryl containing 2 per cent of water should, 
on the basis of Nemst's figures, yield 0.24 cubic centimeter of hydrogen, 
provided the gas was quickly cooled. Hence only a small portion of the 
hydrogen can be attributed to the dissociation of water present in the 

To the question of the importance of ferrous salts in the production of 
hydrogen, it is possible that meteorites, which have usually been regarded 
as free from water, can add testimony of some value. Though it is true 
that in freshly fallen specimens hydrous minerals have not yet been recog- 
nized,^ nevertheless, the researches of Graham, Mallet, Wright, and Dewar, 
besides my analyses of the Allegan, Estacado, and Toluca meteorites, 
have shown that these bodies, when heated, give off much gas, rich in 
hydrogen. If these meteorites really contained no water, either original 
or by absorption from the earth's atmosphere, the hydrogen obtained 
from them can not be attributed to the decomposition of water; it must 
have been held within the mass of each meteorite, either entrapped or 
occluded.^ But in several instances, at least, the investigators have stated 
that a certain quantity of water was driven off, though perhaps this came 
from weathered aerolites. The chemical analysis of the Allegan meteorite, 

1 Nemst, Chem. Central-Blatt, 1905, 2, p. 290. 

» Farrington, Jour, of GeoL, vol. 9 (1901), p. 632. 

s It is to be remembered that a few meteorites have been found to contain hydro- 
carbons, from which hydrogen mieht arise, but the presence of these hydrocarbons from 
inorgamc sources is more remarkable than that of hydrogen itself. 


which was dug up while still hot, gave Stokes 0.25 per cent of water. ^ 
Perhaps this was moisture absorbed from the air by deliquescent com- 
pounds, such as lawrencite; still, on the other hand, there appears no 
reason, at the present time, why a part of this water should not be a pri- 
mary constituent of the meteorite. This uncertainty points out the desir- 
ability of further, and more critical, studies upon the composition and 
properties of meteorites, before attempting to base an argument upon the 
absence of water in these bodies. 

Other possible sources of hydrogen are hydrogen sulphide, hydro- 
carbons, and the products of radioactivity. As the decomposition of 
sulphureted hydrogen has already been mentioned, and is also treated 
under the head of that gas, it need not be discussed here. Hydrocarbons 
can only be represented in small quantities in igneous rocks, and should 
produce more methane than free hydrogen. Unless the analysis shows 
much marsh-gas, hydrogen from this source must be unimportant. 


The carbonates of most metals are decomposed by heat with the liber- 
ation of carbon dioxide. On this account the determination of the carbon 
dioxide yielded by rocks which have undergone much carbonation is of 
little value. Many rocks which appear to be perfectly fresh have neverthe- 
less suffered slight carbonation while in the zone of weathering, and thus 
possess carbon dioxide in a combined state ready to be evolved when suffi- 
ciently heated. This carbon dioxide from the non-gaseous constituents 
of the rock embarrasses the determination of the free gas, since there is no 
way of separating the carbonic acid from these different sources. 

The degree of heat necessary to decompose carbonates throws some 
light on the question. Erdmann and Marchand state that already at 
400^ traces of carbon dioxide are given off from calcium carbonate.' The 
studies of Debray show that at the boiling-points of mercury and sulphur, 
350® and 448° respectively, the development of CO, from calcite in vacuo 
is inappreciable.' The same investigator found that at 860° calcite gives 
up carbonic anhydride until a pressure of 85 millimeters is reached, when the 
action ceases. At 1040° the pressure may rise to 520 millimeters before 
the evolution of gas is stopped. In the presence of carbon dioxide at the 
ordinary atmospheric pressure, calcite retains all of its optical and other 
properties unaltered, even at 1040°. Carbon dioxide from calcium carbonate 
is thus not of any quantitative importance below 450°. In general, most 
of the carbonic acid from the rocks is expelled at temperatures above 
450°. But considerable CO, often appears before the heat reaches 400°, 
as is shown by the Baltimore gneiss. Perhaps this gas may be assigned to 
ferrous carbonate. Iron carbonate would be expected to decompose more 
readily than calcium carbonate, though I have been unable to discover at 
what temperature the process commences. 

» Ante, p. 21. 

' Erdmann and Marchand, cited by Gmelin-Kraut, Anorg. Chem., 2, p. 354. 

' Debray, Gomptes Rendus, vol. 64, p. 603. 



When a finely powdered igneous rock is treated with hydrochloric 
acid and gently warmed, a few small bubbles of carbon dioxide usually 
are seen to rise to the surface of the acid. This gas comes from the action 
of the acid upon small quantities of carbonate present in the rock. To 
test the quantitative importance of this action and to discover whether 
other gases are freed by acid, 25.13 grams of diabase from Nahant, Massa- 
chusetts,^ were placed in a flask connected with the mercury-pump, and the 
air removed. Dilute sulphuric acid was introduced into the flask through 
a dropping funnel. The gas developed in the cold during the fiirst 2} hours 
was found to have the following composition: 

Table 35. 



Hydrogen salphide 

Carbon dioxide 







• • • • 







Practically all of the carbon dioxide thus set free is to be assigned to 
a carbonate. 

The apparatus was allowed to stand for three days, during which time 
more gas came off. At the end of this period, the powder was washed, 
dried, and then submitted to the ordinary process of heating in the tube. 
Of the gas received, 38.19 per cent, or 0.62 volume per volume of rock, 
was carbon dioxide. Powder from the same specimen of diabase, not 
treated with acid, yielded 8.51 volumes of carbonic anhydride in the com- 
bustion-tube. This amounted to 61.25 per cent of the total gas.' 

While carbon dioxide, both gaseous and liquid, occurs in minute cavities 
in certain minerals and rocks, and while rocks also, doubtless, contain 
some of this gas in a state of occlusion, it seems probable, on account of 
the wide dissemination of carbonates in small quantities through the 
accessible rocks near the earth's surface, that the greater part of the carbon 
dioxide obtained by the method of heating rock material in vacuo is derived 
from the decomposition of carbonates in the combustion-tube. It may 
be assumed that more of the carbonates in igneous rocks are secondary 
than primary.'^ But though a knowledge of this immediate source of much 
of the carbon dioxide in the rocks does not lead far toward the elucidation 
of the problem of the ultimate source of this gas, it imposes no restrictions 
upon the more comprehensive view that the carbonic acid which is now 
locked up in the rocks chemically, as a result of weathering and carbona- 
tion, was given to the atmosphere and hydrosphere originally from the 
magmas themselves. 


Metallic iron and ferrous salts reduce carbon dioxide to monoxide 
under practically the same conditions that they liberate hydrogen from 

1 Analysis No. 88. 

' Analysis No. 86. 


3FeO + CO, = FesO^ + CO 

While this action commences below 400^, it takes place slowly, and it is 
chiefly at higher temperatures that it becomes of quantitative importance. 
As in the case of hydrogen and water- vapor, this reaction is reversible, 
the direction in which it will proceed depending upon the proportions of 
the substances present. Either metallic iron or ferric oxide, heated in a 
mixture of equal parts of carbon monoxide and carbon dioxide, produces 
ferrous oxide.^ Siderite at red heat passes into a magnetic oxide ^ith the 
formation of both carbonic acid and carbonic oxide. According to 
Dobereiner this reaction takes place as follows:^ 

SFeCO, = SFeO.FeA + 4C0, + CO 

Glasson,' however, says that 4FeO.Fej08 results, at first giving two parts 
of CO3 and one of CO, but that later the proportion changes to five parts 
of CO3 and one of GO. 

It is, therefore, the normal thing for a rock containing carbon dioxide 
(whether occluded, or in cavities, or a carbonate) and iron in the ferrous 
condition to generate carbon monoxide on the application of heat. In 
this connection it may be noted that carbon monoxide rises very conspic- 
uously in relative importance whenever there is metallic iron present in 
the material tested. The iron-bearing basalt of Ovifak, Greenland, gave 
21.63 per cent of this gas compared with 46.50 per cent of the dioxide,** 
the Allegan meteorite, 38.61 per cent of CO and 41.74 per cent of CO,;* 
while the Estacado meteorite developed 29.31 per cent monoxide and only 
28.47 per cent dioxide.* These were specimens of stony material contain- 
ing grains of metallic iron. Quite different is the Toluca iron meteorite, 
whose nearly pure metal evolved 71.05 per cent carbonic oxide with but 
6.40 per cent carbonic anhydride.' Wright's figures for iron meteorites 
are equally noted for high percentages of carbon monoxide.' 

However, there are two other chemical sources for carbon monoxide, 
one of which is, perhaps, especially applicable to iron meteorites. It is 
known that the carbides of chromium and iron, when heated with the 
oxides of these metals, produce carbonic oxide.* As these meteorites often 
contain considerable carbon, some of it perhaps as a carbide, scrupulous 
care is always necessary in preparing the metal for the analysis, to avoid 
introducing any rust from the oxidized exterior of the mass. 

The other principle must always be operative in the combustion-tube. 
Boudouard has shown that at the temperatures of the combustion-furnace 
hydrogen reduces carbon dioxide, forming carbon monoxide at the expense 
of both hydrogen and the dioxide.' 

» Wright and Luff, Jour. Chem. Soc., vol. 33 (1878), p. 604. 
' Cited by GmeUn-Kraut, Anorg. Chem., vol. 3, p. 319. 
' Analysis No. 45. 

* Analysis No. 106. 
» Analysis No. 107. 

* Analysis No. 108. 
^ See p. 6. 

' Borchers and McMillan, Electric Smelting and Refining, p. 545. 
•O. Boudouard, Chem. Central-Blatt (1901), 1, p. 1360. 


COj+H,=CO + HjO 

Equal volumes of hydrogen and carbon dioxide heated at 850^ for 
one hour gave CO, 44.3 per cent, CO 8.3, H, 42.0, and H^O 5.4 per cent. 
Heated for three hours under the same conditions, the proportion of carbon 
monoxide rose to 18 per cent. When rocks are heated for analysis, the 
gas is usually pumped off at short intervals, and this reaction, because of 
its slowness, becomes less important. Htlttner has appealed to thb reaction 
to explain the presence of carbon monoxide in minerals. 

But metallic iron also has a penchant for absorbing carbon monoxide 
at the proper temperature. This process is usually called occlusion, and 
may perhaps partake of the nature of a combination in which the gas 
temporarily unites with the iron as iron carbonyl, Fe(C0)4,* an unstable 
compound readily giving up carbonic oxide. It seems likely that a portion 
of the carbon monoxide developed from these irons, particularly those of 
meteoritic origin, actually exists in the iron as monoxide, and that not all 
of it has been formed by reduction of the dioxide. 


Certain rocks, when heated, disengaged sulphur dioxide in considerable 
quantities.' These were ferruginous rocks of rusty appearance, generally 
metamorphosed pyritiferous shales which had undergone much weather- 
ing. By oxidation, the original pyrite had been partially converted into 
ferrous sulphate (Fe804) and basic ferric sulphate (Fe^jO,), both of which 
were decomposed by the heat of the combustion-furnace. 

2FeS04 = Fe A + SO, + SO, Fe,S ,0, = FejOi + 280, 

The sidphur trioxide was reduced to the dioxide either by hydrogen sul- 
phide, hydrogen, ferrous oxide, or sidphur. 

SO,+H,S = SOj+HjO + S 

It has been my observation that whenever sulphur dioxide was evolved 
a slight sublimate of sulphur collected toward the cool end of the tube. 
This may have been derived from the reaction above, or from hydrogen 
sulphide and sidphur dioxide, coming from ferrous disulphide and sulphate, 
respectively, and which can not exist together. 

2H,S + SO,=2HjO + 3S 

The sulphur dioxide obtained in the study of rocks is all assigned to 
these reactions, though it is not impossible that this compound may occur 
in small quantities, as a gas or a liquid, imprisoned in minute cavities. 


When iron pjnites (FeS,) is heated in a stream of hydrogen, ferrous 
sulphide (FeS) and free sulphur result." Though no hydrogen sulphide 

^ Fe and CO also exist feebly united in other proportions, as iron pentacarfoonyl, 
Fe(0O)f, and heptacarbonyl, FeCCO)^. 
' Analyses Nos. 43, 65, 93, and 109. 
* Rose, Pogg. Ann., vol. 6, p. 533. 


IS formed in this manner, that gas is produced when pyrite is decomposed 
by steam at high temperatures.^ 

FeSa + HjO = FeO + H,S + S 

As pyrite is frequently present in igneous rocks which generally evolve 
water-vapor upon the application of heat; the limited quantities of hydro- 
gen sulphide obtained may be explained in this way. But unless the 
hydrogen sulphide be removed, this process can proceed only to a certain 
point, for, according to Berzelius, iron disulphide is formed when FeCO,, 
Fcj04, Fe,Os, or Fe(OH), is heated with hydrogen sulphide to temperatures 
between 100® and red heat.' At red heat, a current of dry hydrogen sul- 
phide completely converts Fe,04 into FesS4 in two hours, while a still 
farther increase of temperature results in the formation of FeS and a deposit 
of sulphur.' 

An inspection of the analyses shows that sulphureted hydrogen is 
rarely obtained in large amounts from igneous rocks. An average of 75 
analyses from a wide range of rocks (but omitting bituminous shales) 
gave 0.59 per cent of this gas. But this figure is not a good working aver- 
age, since it has been much influenced by the high sulphide percentage 
of a few individuals. Deducting the five highest of these, the remaining 
70 analyses give an average of 0.27 per cent of hydrogen sulphide. In 19 
cases out of the 75, this gas was entirely lacking. 

While it is probable that not much of this gas was given off from the 
rock material in the first place, a portion of it doubtless disappeared before 
passing through the pump into the gas-receiver. At the temperature of 
the combustion-furnace, hydrogen sulphide is apt to be partially dissociated 
into its elements, thus swelling the already large volume of hydrogen pres- 
ent.^ Gautier states that sulphur heated in a tube filled with hydrogen 
sulphide causes the decomposition of the gas with the result that its sul- 
phur is added to the free sulphur, while hydrogen, nearly pure, remains.' 
When the rock has been considerably weathered, and some of the pyrite 
oxidized into iron sulphate, so that, in addition to hydrogen sulphide, 
sulphur dioxide is disengaged, the former gas will be partially or com- 
pletely decomposed, depending upon the relative proportions of the two 



The bituminous shale from Newsom's Station, near Nashville, Ten- 
nessee,' yielded sulphureted hydrogen to the extent of 30.94 per cent of 
the total gas, which is equivalent to the unusual amount of 29.38 volumes 
of hydrogen sulphide from one volume of shale. A specimen of the well- 

^ With an excess of steam the reaction goes further: 3FeS,+4H,0 « Fefit+SHfi 

' Beraelius, cited by Graham-Otto, Anorg. Chem., 4, p. 718. 

* Sidot, Chem. Gentral-Blatt, vol. 40 (1S69), p. 1038. 

* See p. 47. 

* Gautier, Comptes Rendus, vol. 132 (1901), p. 189. When jpvrite is heated either 
in a vacuum, or in a stream of dry carbon dioxide, FerSg and free sulpnur result (Berzelius, 
Rammekberg^ cited in Gmelin-Kraut, Anorg. Chem., 3, p. 335). 

* Analysis No. 41. 


known ''oil rock" from a lead and zinc mine near Platteville, Wisconsin, 
produced 6.79 per cent hydrogen sulphide, corresponding to 3.90 volumes 
per volume of rock.* How prolific a source of hydrogen sulphide the organic 
matter in certain shales may be, is indicated by these two experiments. 
If a shale of this sort, undergoing extensive metamorphism, did not lose 
all of its sulphur compounds during the transforming process, the meta- 
morphic product might still be distinguished by a high content of hydrogen 
sulphide. Perhaps the specimen of Baltimore gneiss obtained from Spring 
Mill, on the Schuylkill River,* from which 4.91 per cent, or 0.30 volume, 
of sulphureted hydrogen was extracted, may have been derived from such 
a shale. Other sulphur compounds in small amounts have been noted 
in the gases from rocks. The potassium hydroxide solution in the Lunge 
nitrometer, after having absorbed whatever hydrogen sulphide and carbon 
dioxide there may have been in the gas under analysis, frequently emits 
an odor suggesting a mercaptan. When air is let into the pump and tubes, 
after the removal of the gas for analysis, and then pumped out, it usually 
is charged with odors of more or less offensive nature. These suggest 
that other complex reactions prevail at the high temperatures employed 
in extracting the gas. Gautier detected a trace of ammonium sulphocyanide 
in the gas from a granitoid porphyry from Esterel.' 


Moissan believed that the hydrocarbons of the petroleum type which 
occur in the earth's crust were, in many cases, derived from the action of 
water upon metallic carbides in the deep interior.^ His important researches 
upon carbides form the experimental basis for the hypothesis that the 
methane obtained by heating igneous rocks has resulted from these com- 
pounds. Even with cold water, the carbides of barium, strontium, calcium, 
and lithium give pure acetylene, while under the same conditions aluminum 
and beryllium carbides generate pure methane. 

CaC, + 2H,0 « Ca(OH), + C^H, A1,C, + 12H,0 = 4A1(0H), + 3CH, 

The carbides of the rarer metals, cerium, lanthanum, yttrium, and 
thorium, yield various mixtures of acetylene and marsh-gas; from manga- 
nese carbide, marsh-gas and hydrogen result. But the most remarkable 
of the carbides is that of uranium, which with water at ordinary tempera- 
tures produces (in addition to a gaseous mixture of methane, hydrogen, 
and ethylene) both liquid and solid hydrocarbons. Under ordinary condi- 
tions water does not decompose the carbides of molybdenum, tungsten, 
chromium, or iron. 

These reactions suggest two alternative hypotheses to explain the 
occurrence of methane in the gas obtained from igneous rocks. The most 
limited of these supposes the marsh-gas to be produced from a carbide 

' AnalyBis No. 42. 
' Analysis No. 28. 

* Gautier, Comptes Rendus, vol. 132, pp. 61-62. 

* Moissan, Proc. Roy. Soc., vol. 60 (1897), pp. 166-160. 


in the combustion-tube. Such a carbide must have withstood the action 
of water, both magmatic and meteoric, ever since the solidification of the 
rock. The other hypothesis seeks to avoid this difficulty by postulating 
carbides in the very hot rocks where the hydrogen and oxygen may not be 
combined as water; then, at a later stage, it allows water to decompose 
the carbides wdth the evolution of marsh-gas, which is retained within the 
rock. In this case the gas itself would exist in the rock specimen tested. 

It is to be noted that several of these carbides, including that of the 
widespread element calcium and the less stable sodium and potassium 
compounds,^ give acetylene when decomposed by water. In none of the 
rocks examined, with one exception, has acetylene been detected. This 
may possibly eliminate calcium carbide from the gas-contributing com- 
pounds in the rocks. The absence of acetylene also carries with it some 
slight evidence against carbides in general, since calcium plays a very 
important r61e in rock evolution, and it is not likely that acetylene, if 
formed, would pass into methane. However, it is not impossible that 
aluminum carbide, which yields methane with water, may exist in the 
earth's crust while calcium carbide is lacking. According to F. W. Clarke,' 
aluminum constitutes 8.16 per cent of the solid crust of the earth, while 
iron and calcium comprise 4.64 and 3.50 per cent, respectively. Aluminum 
also forms very stable compounds in nature. Moreover, aluminum oxide 
fused in the electric furnace with calcium carbide gives yellow crystals of 
aluminum carbide.' Perhaps at high temperatures iron carbide might 
be decomposed by steam with the formation of marsh-gas. 

Besides carbides, organic matter suggests itself as a possible source of 
the methane. This organic matter may have been either (1) accidentally 
introduced into the combustion-tube, or (2) have been incorporated in the 
rocks from life which inhabited the earth during the later stages of growth, 
as outlined by the planetesimal hypothesis. The first possibility may be 
practically dismissed, since great care was exercised to avoid the intro- 
duction of any foreign matter with the rock powder. If dependent upon 
such accidental conditions, this gas would only occasionally be present. 
Under the planetesimal hypothesis, life may have existed long before the 
growth of the planet was completed and its present size attained. Organic 
deposits buried in sedimentary beds which have since undergone exten- 
sive metamorphism should furnish marsh-gas. These rocks, worked over 
and reworked by the volcanic activity in Archean times, might perhaps 
account for the widespread occurrence of this gas. Formed in this way, 
it may be retained in the rocks as a free or occluded gas, since it is very 
stable at high temperatures. 

Other theoretical sources of methane are high-temperature reactions 
within the combustion-tube, in which hydrogen and the oxides of carbon 
participate. Brodie produced 6 per cent of marsh-gas by submitting 

* Moiflsan, Jour. Chem. Soc., vol. 64, 2, p. 332. 

» F. W. Clarke, Bull. 168 U. S. G. S. (1900), p. 15. 

* Moissan, Gomptes Rendus, 125 (1897), pp. 839-^4; Jour. Chem. Soc., vol. 64 
(1898), 2, p. 161. 


approximately equal volumes of hydrogen and carbon monoxide to the 
action of electricity for five hours in an induction-tube.* From his results 
he expressed this reaction by the equation, 


Though it is not safe to assume that this reaction will take place at 
the temperatures of the combustion-furnace, the observation that marsh-gas 
is obtained when a rock powder, exhausted of its gases, is exposed to the 
air for a few months, and reheated, possibly points toward some reaction 
of this nature. 


If the nitrogen which is obtained from heating igneous rock powders 
in vacuo is derived from some chemical compounds decomposable at red 
heat, a metallic nitride at once suggests itself as the most probable form in 
which the nitrogen would occur. Iron nitride may be taken as the type for 
discussion. While different nitrides of iron, having the compositions of 
FejNj, FejN, and FejNj, have been described by some authors, a compre- 
hensive study, by Fowler, of this formerly little-known compoimd, forces 
the conclusion that there exists ' only one iron nitride, FesN. This com- 
pound may be prepared by the action of ammonia either upon ferrous 
chloride, or finely divided iron. While the action between ferrous chloride 
and ammonia commences near the melting-point of lead (327°), a tem- 
perature of 600° appears to be necessary for the production of iron nitride 
in quantity.' The nitride can also be produced at 850° to 900°, but this 
is probably the highest limit of the reaction.* 

This nitride is very soluble in dilute acids, giving ammonia. When 
heated to redness in hydrogen, ammonia results. At 200° it is oxidized 
in the air to ferric oxide, abandoning nitrogen, which does not appear to 
be oxidized. At 100° steam causes a slight evolution of ammonia. Accord- 
ing to Fowler the temperature of decomposition of iron nitride in an inert 
gas (nitrogen) must certainly be above 600°. 

Silvestri has found iron nitride coating some of the fumarole deposits 
of Etna,* and Boussingault * recognized nitrogen in the Lenarto meteor- 
ite by certain tests which led him to believe that it existed there as a 
metallic nitride. Silvestri discovered that at red heat the nitride from 
Etna was decomposed, delivering up its nitrogen. His experiments show- 
ing that iron nitride may be prepared artificially, by igniting the ordinary 
lava in a current of ammonium chloride vapor, probably illustrate what 
takes place in the fumaroles. But this nitride, which derives its nitrogen 
from ammonia or ammonium salts, in no way requires the existence of 
nitrides within the magma itself, except as a possible source for the nitro- 
gen which unites with hydrogen to form the ammonia. There is danger 
in transferring the characteristics of fumarole deposits, which are formed 

' Sir B. C. Brodie, Proc. Roy. Soc., vol. 21 (1873), p. 246. 
» Fowler, Jour. Chem. Soc, vol. 79 (1901), pp. 285-299. 
» Fowler, Chem. News, vol. 82 (1900), p. 246. 

* Beilby and Henderson, Jour. Chem. Soc., vol. 79, p. 1246. 
» Silvestri, Pogg. Ann., vol. 167 (1876), pp. 166-172. 

* Boussingault, Comptes Rendus, vol. 63 (1861), pp. 78-79. 


in limited quantities under the quite exceptional conditions of abundant 
currents of free gases and very active hot vapors, to the main magmas. 

If the nitrogen obtained from rock powders be derived from a nitride, 
it should be accompanied by ammonia, since, in the presence of hydrogen 
or water-vapor, it is this gas, rather than free nitrogen, which is given off. 
Tests made with Nessler's solution show that ammonia is one of the gases 
extracted from rocks, though always appearing in limited amounts. In 
the process ordinarily employed for extracting the gas, it is absorbed by 
the calcium chloride drying-tube. Ammonia is scarcely to be considered 
as a source of free nitrogen, since this compound is only dissociated at the 
temperature of the electric spark. 

Whether all of the free nitrogen can be assigned to the decomposition 
of iron nitride may be tested with the quartz from New South Wales.^ Sup- 
posing all of the iron in this quartz to have existed as iron nitride, and 
to have been completely decomposed without the production of any am- 
monia, the analysis still shows an excess of nitrogen over what could have 
been produced in this way. The reaction may be taken as 

2Fe,N = 4Fe + N, 

102.72 gmB. quarti contained 0.0058 gm. Fe 

F^ (as PeaN) required to give 1 c.c. nitrogen 0100 gm. 

Nitrogen poonble from reaction 58 c.c. 

introgen actually obtained (0^ and 760 mm.) 86 o.o. 

Ezoeaa of nitrogen 28 o.o. 

A duplicate determination of the iron in this weight of quartz gave only 
0.0048 gram; on this basis, the excess of nitrogen would be still greater. 
It is highly improbable that all of the iron in this quartz was combined as 
a nitride. Some of it was unquestionably pyrite. To ascertain how much 
of this nitrogen can be ascribed to atmospheric air adhering to the tubes, 
as well as to leakage during the process of extraction, a blank combustion 
was resorted to. The empty combustion-tube was kept at bright-yellow 
heat for the length of time which was required to expel the gas from the 
quartz. 0.15 cubic centimeter of gas was collected in the receiver when 
the tube was exhausted by the pump. Adhesion of air to the quartz itself 
might be supposed to increase this figure, though the material used for this 
analysis was not the usual fine powder, but small fragments which would 
be less liable to entrap air. In general, while iron nitride is to be accepted 
as a possible source of this nitrogen, it is inadequate to produce the quanti- 
ties of this gas determined by analysis. The presence of other metallic 
nitrides in this comparatively pure quartz does not seem likely. Silicon 
nitride, however, may be present and might possibly contribute a portion 
of the nitrogen. 


Though occlusion is a phenomenon but imperfectly understood, there 
appear to be three different ways in which it is manifested. In the first 
of these the absorption seems to be dependent upon porosity. An example 
of this is charcoal, one variety of which absorbs 172 volumes of ammonia, 

^ AnalysiB No. 100. 


165 volumes of hydrogen chloride; 97 volumes of carbon dioxide, and 2 
volumes of hydrogen.* The aflBnity of molten silver for oxygen illustrates 
another phase of absorption often classed as occlusion. It has long been 
known that silver absorbs 22 times its own volume of oxygen when melted, 
but gives up most of this gas, often with violence, as it solidifies.' This 
is properly a solution of a gas in a liquid, and not in a solid, as in the case 
of true occlusion. The third type is the absorption of gases by compact 
metals, either on their surface or within their mass, such as the occlusion 
of hydrogen by palladium, platinum, and iron. This is, in the main, inde- 
pendent of porosity. 

Hydrogen is absorbed by these metals at ordinary temperatures, but 
is only given off at higher temperatures. This principle was demonstrated 
by Graham, who placed a thin plate of palladium, charged with hydrogen, 
in a vacuum and observed that at the end of two months the vacuum was 
still perfect. No hydrogen had vaporized in the cold, but on the applica- 
tion of a heat of 100^ and upwards, 333 volumes of gas were evolved from 
the metal.' The degree of heat required to expel hydrogen absorbed by 
platinum and iron was found to be little short of redness, although the 
gas had entered the metal at a low temperature. Another series of experi- 
ments by the same investigator showed that, to be occluded by palladium 
and even by iron, hydrogen does not need to be applied under sensible 
pressure, but on the contrary, when highly rarefied, it is still freely absorbed 
by these metals. These results have been confirmed by Mond, Ramsay, 
and Shields,^ who found that platinum black at very low pressures absorbed 
a certain quantity of hydrogen. On increasing the pressure of the hydro- 
gen up to about 200 to 300 millimeters, a further quantity was absorbed, 
but beyond this point an increase of pressure had comparatively little 
effect. These investigators regarded 1 10 volumes as the amount of hydro- 
gen really occluded by platinum black, although 310 volumes were actually 

Experiments indicate that the quantity of hydrogen occluded depends 
greatly upon the condition of the metal. When chemically reduced, cobalt 
may occlude 59 to 153 volumes, nickel 17 to 18, and iron 9 to 19 volumes.* 
Though common iron wire occludes only 0.46 volume of hydrogen,* this 
same metal, ^when electrolytically deposited, may absorb nearly 250 vol- 
umes of this gas.' The maximum quantity of hydrogen occluded by any 
metal, so far as recorded, is 982 volumes absorbed by freshly precipitated 
palladium.* Dumas has shown that aluminum heated in vacuo to 1400° 
gives off more than its own volume of gas, consisting chiefly of hydrogen 
with a little carbon monoxide, but without traces of carbon dioxide, oxygen, 
or nitrogen.* Under the same conditions, magnesium rapidly expels 1.5 

Barker, Textbook of Physics, p. 183. 

Chimie Min^rale, Moissan, t. 1, p. 203. 

Graham, Chemical and Physical Researches, pp. 283-290. 

Proc. Roy. Soc., vol. 68 (1895), pp. 24^24S. 

Chimie Min^rale, Moissan. t. l,p. 51. 

Graham, Chemicied and Pnysical Researches, p. 279. 

Cailletet, Llnstitut, Nouv. S6r., Ann. 3, p. 44. 

Graham, Chemical and Physical Researchas, p. 287. 

Dimias, Comptes Rendus, 90 (1880), p. 1027. 



volumes of nearly pure hydrogen. Many other metals behave similarly. 
Non-metallic substances appear to possess this property in a lesser degree. 
Porcelain occludes hydrogen, whether because of its porosity or solvent 
qualities is not certain. Quartz is said to be penetrable, at high tempera- 
tures, by the gases from the oxyhydrogen flame,^ which points towards a 
form of occlusion. 

In addition to hydrogen, other gases are occluded. Litharge, when in 
the molten condition, dissolves hydrogen, carbon monoxide, and nitrogen, 
of which it retains a portion on solidifying.' Cast iron, on cooling, retains 
4.15 volumes of carbon monoxide,' which perhaps may be due to the for- 
mation of iron carbonyl, Fe(G0)4, or similar unstable compounds. 

Analyses show that whenever metallic iron is present in notable quan- 
tities carbonic oxide becomes an important constituent of the gas evolved. 
The following analyses of the gases from various types of iron indicate 
the proportions of hydrogen, carbon monoxide, carbon dioxide, and nitro- 
gen which this metal may absorb, given in percentages of the total gas 


Tablb 36. 


White, carhonaceous, cast iron 

Mild steel 

Ordinary gray charcoal iron 

Gray coke iron 


Bessemer steel before adding spiegel — 

Bessemer steel after adding spiegel 

Open-hearth steel 

Capola pig iron 

Horseshoe nail, heated 2 hours 

Same, heated 2 hoars more 


Troost & 


Troost & 

Muller « 











24.3 • 












• • • • 











i Cited by Cohen, Meteoiltenkunde, p. 181. 

iCited by Lane, Bull. Geol. Soc., yol. 5 (1894), p. 264. 

Whatever may prove to be the ultimate significance of occlusion, and in 
whatever condition these gases are stored in the iron, whether it be in the 
nature of a solution, as Mendel6ef has suggested, or as definite compounds — 
hydrides, nitrides, and carbonyls — the fact remains that these gases exist 
within the metal and are in many respects similar to the gases locked up in 
iron meteorites. Fresh iron borings from the interior of a metallic meteorite 
have usually been assumed to be free from any hydration or carbonation 
from terrestrial agencies, and so have been held to contain true meteoritic 
gases. Some question respecting this belief has arisen from certain analyses 
which, as we have seen, indicate secondary action.^ In addition to this 
the gases actually received in the laboratory may not represent the original 

^ Poynting and Thomson, Properties of Matter, p. 204. 

' Le Blanc and Cailletet, cited by Violle, Cours de Physique, t. 1, p. 922. 

* Daniell, Princii>le8 of Physics, p. 327. 

* Experiments with the Toluca iron ; Analysis No. 108. 


proportions on account of the reducing action of iron on carbon dioxide 
and water- vapor. Meteorites of the stony type, unless absolutely fresh, 
are more open to the suspicion of terrestrial hydration and carbonation. 
But the Allegan meteorite gathered up, still hot, within five minutes of 
its fall, has not been subjected to outdoor exposure, though it may have 
absorbed a small amount of moisture and carbonic acid from the atmo- 
sphere since being placed in the National Museum. It yielded somewhat 
more than half of its own volume of gas.^ Fresh material from the interior 
of the E^tacado, Texas, meteorite, heated in a vacuum in the presence of 
phosphorus pentoxide for five hours at 150°, and then allowed to remain 
untouched for several days to enable the drying agent to take up the last 
traces of moisture in the tubes, still yielded at red heat 0.86 volume of gas, 
of which 36.25 per cent was hydrogen.* 

These gases from stony meteorites resemble those from some igneous 
rocks. That this correspondence should exist, is entirely in accordance 
with the view that meteorites have been derived from the disruption of 
small planetary bodies of the nature of the asteroids. As in the meteorites, 
so in the rocks, that portion of the gases which can not have been produced 
by chemical reactions at elevated temperatures, nor from the bursting 
of rock-bound cavities, may fairly be assigned to occlusion. The computa- 
tions indicating the excess of hydrogen obtained from quartz and beryl ' 
over that which might have arisen from the interaction of iron and water 
under the most generous assumptions show that, in some cases, more gas 
may arise from a state of occlusion than from ordinary chemical action. 
The amoimt of occluded gases may be actually greater than that indicated 
by demonstrating the inadequacy of other modes of holding gas. But in 
basic rocks containing hydrated minerals and an abundance of ferrous 
salts, the resulting volumes of hydrogen must doubtless come largely from 
the decomposition of the water of constitution, and the amount of occluded 
gases, if any, is beyond determination by these methods. 

The gases argon and helium, which, according to current chemical 
views, do not form compounds, must exist within rocks either mechanically 
entrapped or in a state of occlusion. There are those, notably Ramsay 
and Travers, who believe in the combining properties of argon and helium; 
but the balance of opinion seems to be on the other side, so far as ordinary 
terrestrial conditions are concerned. Lord Rayleigh concludes his paper 
on the inactivity of these two gases, with this sentence: ''There is, there- 
fore, every reason to believe that the elements, helium and argon, are non- 
valent, that is, are incapable of forming compoimds."^ 

As the chemists' supply of helium comes from certain minerals, chiefly 
those containing compounds of uranium, its occurrence in rocks is a well- 
known fact. Recent studies have revealed the existence of helium in beryl.* 
Argon is perhaps more widely distributed than helium, Gautier having 
detected this element in ordinary granite. The waters of many springs 

^ AnalyBis No. 106. 

' Analysis No. 107. 

» Ante, pp. 46-48. 

* Lord Kayleigh, Proc. Roy. Soc., vol. 60, p. 56. 

» R. J. Strutt, Nature, Feb. 21, 1907, p. 390. 


bring up both helium and argon, proving the presence of considerable 
quantities of these elements within the earth. This rather wide distri- 
bution, when taken in connection with their supposed chemical inertness, 
strengthens the presumption that occlusion, or some form of gas diffusion, 
is prevalent in rocks. Though of much interest to chemists and physicists, 
these gases, on account of their comparative scarcity, do not play a very 
important rAle in general geological problems. Their presence in small 
quantities within the rocks of the earth's crust being established, quan- 
titative determinations become of less value. In the analyses made for 
this paper, whose prime purpose was to determine the range and distribu- 
tion of the common gases, the separation of helium and argon from nitrogen 
was not usually attempted. These gases when present are included in 
the figures given for nitrogen. In the case of pitchblende and carnotite, 
however, helium was so important a constituent of the gas that its pro- 
portions were determined. Carnotite produced 1.28 per cent of helium, 
amounting to 0.04 volume, while pitchblende gave 38.48 per cent, or 0.37 
volume.^ In both of these cases nitrogen also was abnormally high. 

In general, therefore, helium and argon, together with at least as much 
of the other gases as can be shown not to have been produced by chemical 
reactions or the biursting of inclosing walls, are to be attributed to occlusion 
or some form of diffusion not distinguishable from occlusion. In many, 
and perhaps most, rocks this will not be the major part, for, of the three 
gas-liberating processes, that by chemical interaction imder the influence 
of heat appears to be the dominating one. 



While chemical reactions and the phenomena of occlusion imply that 
gis exists in the interior of the earth, the presence of gas inclosed in ca\d- 
ties under great pressure adds the further implication that the gas often 
ezeeeded the point of saturation of the magma, at least at the stage of 
solidification. Cavity gases are most abundant in minerals of poorly 
developed cleavage, pointing perhaps towards a strong tendency to escape 
along cleavage planes during, or after, crystallization. The gas inclusions 
in quartz may, however, owe their abimdance not so much to the absence 
of cleavage as to the fact that quartz is generally the last mineral to crystal- 
lise out of a magma, and hence such absorbed gases as did not enter into 
the other crystals would become concentrated in the siliceous residue and 
might supersaturate it. 

It is possibly this freely-moving gas above the point of saturation which 
contributes most to the mobility of lavas. Dissolved gases and vapors, 
while favoring fluidity, would seem to be relatively less effective. But 
the foregoing investigations imply that gases mechanically entrapped in 
crystalline rocks are not very abundant, and suggest that perhaps the 
theory of liquidity due to gas is overworked. On the other hand, it is 
true that as the lava cooled down to the point where the last mineral crys- 

* Analyses 93 and 94. 


tallized, its gas-solvent powers would be increasing, allowing some of the 
gas to pass into solution. At the same time free gas might be occluded by 
the growing crystals. The experiments upon the reabsorption of gas by ex- 
hausted rock powder indicate that a portion of the gas unites chemically as 
the heat diminishes. Because of these processes, liquid lavas may be sup- 
plied with free gas, even when the solidified rocks retain but little free gas. 

As the imprisoned carbon dioxide frequently remains in the liquid 
form up to the critical point (30.9^ C), it must be subjected to a pressure 
of at least 73 atmospheres, which is the critical pressure of this gas. Since 
a pressure of 73 atmospheres corresponds to a column of water 2,470 feet 
in height, quartz cnrstals formed from aqueous solution, under hydrostatic 
pressure simply, can not contain liquid carbon dioxide up to 30.9^ unless 
developed at depths exceeding 2,470 feet. It is to be recognized, however, 
that such crystals might be formed at lesser depths if mechanical pressure 
operated with hydrostatic pressure or replaced it. 

If the quartz crystallized from a lava, say at 1100^ C, the effect of 
cooling down to ordinary temperatures upon both the size of the cavity 
and the pressure of the inclosed carbon dioxide must be taken into account. 
If we take the case of a cavity found to be entirely filled with carbon dioxide 
at the critical point (30.9^ C. and 73 atmospheres), it is possible, by the 
use of Van der WaaPs equation, to calculate the pressure to which the gas 
would be subjected if the quartz were heated to 1100^. This pressure is 
found to be 756 atmospheres,^ provided the size of the cavity remains 
constant. But as most minerals contract on cooling, the volume of the 
cavity diminishes at the same rate as though it were filled with the material 
of the inclosing walls.' The coefficient of expansion of quartz is given as 
0.00003618. Assuming for the sake of simplicity that the rate of expansion 
does not vary greatly with changing temperatures,' quartz, cooling from 
1100® to 31®, would contract to an extent of about 3.87 per cent of its 
original volume. Since the contraction of the quartz diminishes the size 
of the cavity and increases the pressure by 3.87 per cent, the original 
pressure need be only 727 atmospheres, which corresponds to the pressure 
beneath 9,100 feet of average rock. To fill cavities forming in crystals at 
1100® with carbon dioxide which is so condensed that it will pass into the 
liquid state just at the critical temperature when the rock cools down, 
a preaaure corresponding to a depth of at least 9,100 feet, or its mechanical 
equivalent, would seem to be required. If, when warmed under the micro- 
scope, the liquid carbon dioxide is found to pass into the gaseous state at 
temperatures below 30.9®, and the cavity contains only carbon dioxide, 
or carbon dioxide and water, these must have been entrapped under a 
pressure less than 727 atmospheres, or else the crystal was formed at a 
temperature above 1100®. 

* By starting with the equation p « ~Z/h ~ ^ ** ^^® critical point where the values 

of the constants are taken as i?»0.003684: a«0.00874; &-0.0029; v»36->0.00S7, 
and substituting for the critical temperature, r« 1100^+273^, the theoretical value of 756 
atmospheres for the pressure at 1100^ is obtained. 

^Daniell. Principles of Phvsics, p. 370. 

' It would, however, slowly increase with the increase of temperature. 


The estimate of 9,100 feet for the minimum depth (where weight alone 
acts) at which igneous quartz crystals now containing carbon dioxide, 
liquid up to 30.9^, could have been formed/ applies best to those cases in 
which only carbon dioxide exists in the crystal cavities. If there are other 
gases and liquids present in appreciable quantities, this figure becomes less 
applicable, since the constants a (denoting an internal force or attraction) 
and b (representing the sum of the spheres of influence of all the molecules 
in the space i') used in Van der Waal's equation are not the same for 
all gases. At how much greater depths than this the crystallization of cer- 
tain specimens of quartz actually did take place, if rock weight alone was 
involved, may, perhaps, be estimated by a painstaking determination of 
the pressure under which the imprisoned carbon dioxide exists in these 
minute cavities. This might be accomplished by piercing one of the larger 
cavities while submerged in mercury or other liquid, and noting the expan- 
sion of the freed bubble, as first suggested by Sir Humphry Davy. 

Though naturally subject to limitations, it is nevertheless possible to 
throw considerable light upon the nature of cavity inclusions by the use 
of the microscope. Some of the conditions may be stated: 

(1) If, at slightly under 30.9°, the cavity is entirely filled with a liquid 
which completely vaporizes at 30.9°, it contains only carbon dioxide. 

(2) If, at slightly under 30.9°, the cavity is filled with two immiscible 
liquids, one of which passes into the gaseous state at 30.9°, the liquids are 
probably water and carbon dioxide. 

(3) If the cavity, when just below 30.9°, contains a liquid and an appre- 
ciable gas-bubble, and the liquid does not disappear when the slide is 
warmed above 30.9°, the liquid is probably water, and the bubble water- 
vapor with perhaps some of the difficultly liquefiable gases, such as hydrogen, 
nitrogen, or methane. 

(4) If , as is often the case, the temperature at which the liquid in a 
cavity disappears is found to be several degrees below the critical tempera- 
ture of carbon dioxide, two interpretations are possible: either the carbon 
dioxide is subject to a pressure less than 73 atmospheres, or else there is a 
small proportion of another less liquefiable gas present. If the cavity be 
opened and only carbon dioxide be found, the pressure under which the 
gas existed, and from that something as to the conditions under which 
the crystal was formed, can be computed from the temperature at which 
the liquid disappeared. If another gas, such as hydrogen or nitrogen, be 
found and identified, it is possible, by using an equation,' to calculate the 
relative proportions of the two gases from the critical temperature of the 
mixture. Thus a cavity containing a mixture of carbon dioxide and 
nitrogen which had a critical temperature of 29° would hold 98.7 per cent 

* This figure is based on the assumption that the quarts crystallized at 1100^; if it 
Is desired to use other temperatures, they can be substituted in Van der Waal's equation 
and the corresponding pressures computed. 

1 ^ « W| + (l(X)--n;<, ^here t is the observed critical temperature of the mixture, t, 
and tt are the theoretical critical temperatures of the two liquefied gases. Then n equals 
the proportion by weight of the first liquid and 100 ~n equals the proportion by weight 
of toe second liquid. 


of the former and 1.3 per cent of the latter. A critical temperature of 28° 
would indicate 98.1 per cent of liquid carbon dioxide and 1.9 per cent of 
liquid nitrogen, while 27** would mean 97.5 per cent of the dioxide and 2.5 
per cent nitrogen. The figures for carbon dioxide and hydrogen are of the 
same general order. In the estimates of the depths at which cavity-bearing 
crystals were formed, made by different methods,* it has been usual to 
assume that only the pressure arising from the weight of the overlying rock 
was involved. 

Sorby examined those cavities which contained only water, or a saline 
solution, and a vacuole left by the contraction of the liquid, as a result of 
the lowering of the temperature- By noting the relative size of the bubble 
and the volume of the liquid, he estimated the temperature to which the 
mineral would have to be heated for the liquid to completely fill the cavity, 
and from this, together with the elastic force of the water-vapor, he com- 
puted the necessary existing pressure in feet of rock. The highest tempera- 
ture found by this method was only 356® C, at which point Sorby belie ved 
that the trachyte of Ponza solidified, while the lowest temperature was 
89® C, obtained from a study of the main mass of granite at Aberdeen. 
But Sorby considered it more probable that the granite crystallized at 
about the same temperature as the trachyte and, assuming that the solidifi- 
cation took place at 360®, he computed that the granite of Aberdeen was 
formed under a pressure of 78,000 feet of rock.' These estimates are based 
upon the unwarranted supposition that when the crystals were formed 
the volume of liquid water was such as to just fill the cavities, and that in 
each case a meniscus at once appeared with a loss of heat. He overlooked 
the fact that the meniscus could not appear until the water reached the 
liqmd condition, no matter at what temperature the growing crystal sur- 
rounded the vesicle of highly compressed water-gas. 

The highest temperature at which a vacuole of this sort can appear 
must, therefore, be the critical temperature for water, or 365® C. In order 
to study this problem, we may, perhaps, best take the special case in which 
the inclosed water passed through its critical state (at 365® and 200.5 
atmospheres pressure) during the cooling of the crystal. The vesicle formed 
in this case may be termed the critical vacuole. It may be assumed that 
the growing crystal inclosed the water at some temperature in the neigh- 
borhood of 1100®, which is an average temperature for the solidification of 
lavas. Starting thus with a cavity formed at 1100®, in order to allow the 
water on cooling to pass through the critical state, an original pressure 
of 1,070 atmospheres is necessary according to Van der Waal's equation,' 
provided the size of the cavity remained constant. But if 2.66 per cent is 
allowed for the shrinkage of the cavity while cooling down^ from 1100® 

^ See Geikie's Textbook of Geology, vol. 1, pp. 144-145. 
' Sorby, Quart. Jour. Geol. Soc. London, vol. 14, p. 494. 

■ p— — r — =-. In this case the values of the constants for the critical point 

were taken to be ««.003607: a-.01173; 6-.00151; V-36-.00453. By substituting 
for the critical temperature, r« 1100 +273, the equation gives the theoretical value of 
1,070 atmospheres. 
♦ Ante, p. 62. 


to 305^, the original pressure need be only approximately 1,040 atmospheres, 
a preBBure which corresponds to 13,000 feet of rock approximately. 

In this critical case the meniscus appears as soon as the temperature 
falls below 365^. Since pressure exerts but little influence on the volume 
of liquids, the shrinkage of the water in the cavity, and hence the growth of 
the gas-bubble, is largely a function of the fall in the temperature, and, with 
a knowledge of the varying coefficient of expansion, the relation between the 
sise of the bubble and the volume of the liquid could be computed for any 
temperature. The correction for the constriction of the cavity between 
365^ and 20^ amounts to a little more than one per cent, which is to be 
added to the size of both the cavity and the vacuole in computation. 

If these principles be true, a vapor-bubble relatively smaller than the 
critical vacuole may be interpreted to mean that the meniscus did not appear 
in the cavity until the crystal had cooled below the critical temperature, 
t. e., that at this temperature the water was more than normally condensed, 
owing to a pressure exceeding the critical pressure. On the other hand, 
a vapor-bubble relatively larger than the critical vacuole means that, 
although it did not appear until below 365^, it began as a sizable vesicle 
when it did start, owing to tbe lower pressure and more rarefied condition 
of the water-gas. 

On the basis of his experiments, Sorby estimated that a vesicle amoimt- 
ing to 28 per cent of the volume of the liquid in the cavity would vanish 
when the water was heated to 340® C. According to this figure, a vacuole 
occupj^g in the neighborhood of 30 or 35 per cent of the volume of the 
liquid should correspond to a shrinkage of the water from the critical 
point to ordinary temperatures. But this figure has not been confirmed 
by other investigators. Unfortimately the figiures obtainable for the ex- 
pansion of water up to the critical point vary within such wide limits that 
it does not seem advisable at the present time to attempt to calculate the 
rdative dzes of the critical vacuole and the inclosing liquid. 

The difficulties involved in applying these principles are considerable. 
SSrkel has pointed out that, even in cavities within the same crystal, there 
is much variation in the relative volume of the vapor-bubble and the 
liquid, from which the inference is drawn that the vapor-bubbles are due 
to causes other than contraction on cooling.^ Before this conclusion can 
be accepted with confidence, due consideration must be given to the loca- 
tion of the cavities within the crystal, and also to the evidence that they 
are all primary inclusions. In an ascending lava subject to a steadily 
diminishing pressure, those cavities formed during the early stages of 
crystallization may be developed under conditions quite different from the 
cavities later inclosed in the outer parts of the crystals. If systematic 
differences in the cavities can be foimd to correspond with variations in 
thdr location, something might be learned of the history of the lava during 
the period of crystallization. Secondary fluid inclusions, formed subse- 
quent to the solidification of the magma, must obviously be recognized 
and avoided, whenever possible, in attempting to estimate the conditions 
under which crystallization took place. 

^ Zirkel, cited by Geikie, Textbook of Geology, 1, p. 145. 


A diflSculty of a more serious nature, apparently, suggested by Professor 
Iddings, lies in the change of volume of the magma in the passage from the 
liquid to the crystalline form. Some magmas, such as those of granitic 
rocks, contract so appreciably upon crystallization that it is conceivable 
that the last crystals to form, those of quartz (which also contain the most 
liquid and gas inclusions) might crystallize under reduced pressures in 
spaces inclosed by crystals of the minerals already formed. The relative 
size of the bubble of vapor in the cavity and the accompanying liquid 
would, in such cases, not correspond directly to the depth beneath the 
surface at which crystallization took place, even when nothing but hydro- 
static pressure affected the lava column. 

In the present defective state of knowledge as to the modes and condi- 
tions which obtain in lavas penetrating the shell of the earth, it is by no 
means safe to assume that the pressures to which an igneous intrusion is 
subject are merely those represented by the overl3ring rock or a lava 
column reaching to the surface. An ascending tongue of lava may extend 
to great depths and be affected by pressures brought to bear upon it in its 
lower part, which might be in excess of those represented by the depth of 
the head of the column, to an unknown degree. So also it is possible that 
lavas may become involved in mechanical deformations and thus be subject 
to special pressures in no close correspondence to their depth. 


The reversible reactions involving hydrogen, water, and iron com- 
pounds, which cause imcertainties in the extraction of gases by heat, 
are also operative within the earth. In the laboratory, when either ferrous 
salts and water, or ferric compounds and hydrogen, are heated in tubes 
without the removal of the products, reversible reactions set in until a 
condition of equilibrium is established. Hydrogen and water, ferrous and 
ferric salts are all present in a state of balance. In the interior of the earth 
the heated, though solid, rocks should, it would seem, behave similarly, 
though hindered by the slowness of diffusion. Nor should liquid magmas 
constitute any exception to the law. Both hydrogen and water-gas, theo- 
retically, should be present in liquid magmas and heated solid rocks. The 
chief uncertain factors are high temperatures, and pressures. 

The effect of pressure on chemical equilibrium is to favor the formation 
of that system which occupies the smaller volume, but if there is no change 
in volume, in passing from one system to the other, the increase of pressin-e 
presumably has no influence on equilibrium.^ In the reaction 

3FeO+HaO :!l^ Fe,04+H, 

considered as a thermochemical equation, the number of gaseous mole- 
cules, and hence the volume of gas, always remains the same, so that it 
is not likely that this action will be influenced by change of pressure. A 
rise of temperature favors the formation of that system which absorbs 
heat when it is formed.' A comparison of the amoimt of heat liberated by 
oxidizing three molecules of FeO to Fe,04 and one molecule of H, to HjO 

^ Jones, Physical ChemistiT. p. 514. 

' Van't Hoff, Lectures on Tneoretical and Phys. Chem., Pt. 1, pp. 161-164. 


shows that, in the former case, 73;700 calories are evolved, and in the 
latter 68,300; that is, 3FeO+H,0 — ► Fe,O4+H, + 15,400 calories. As 
heat is evolved in this process, a rise of temperature would accelerate the 
reaction in this direction less than the reverse. In other words, the higher 
the temperature, the more would the formation of ferrous oxide and water 
be favored as compared with the conditions at lower temperatures. 

Be<^use of this, there is much reason to suppose that, at the depths 
where lavas originate, hydrogen and oxygen exist combined as water, 
since up to temperatures of 2000^ C, the dissociation of water takes place 
only to a limited extent. If a state of equilibrium between hydrogen, 
water, and the iron compoimds were established in the heated interior 
where a magma originated, as soon as it commenced its way upward and 
began to lose heat the condition of equilibrium would be destroyed. With 
the falling temperature, the tendency to re^tablish equilibrium would 
favor the formation of that system which was produced with the libera- 
tion of heat, i. e., magnetic oxide and free hydrogen. In ascending lavas 
which are losing heat, the tendency, therefore, is to produce hydrogen and 
magnetite, or ferroso-ferric compoimds. This is doubtless an important 
source for the hydrogen which is so copiously exhaled during a volcanic 
eruption. At the same time, this process accoimts for the widespread 
occurrence of magnetite in igneous rocks. The considerable deposits of 
magnetite, formed apparently from magmatic segregation, which are com- 
mon in various regions, may, perhaps, owe their origin to a combination of 
causes, in which this equilibrium reaction is an important factor. 

In general, these reversible reactions tend to show that it is but a short 
step from hydrogen to water, and from carbon dioxide to monoxide, and 
mee versa, and that all of these must occur within the earth owing to the 
processes tending toward equilibrium. Whether hydrogen, in a particular 
ease, occurs in the magmas in the free state, or in the form of water-gas, 
therefore becomes relatively unimportant. Because of this variation of 
state, the problem becomes more complex and broader in scope. For the 
most part, these water-gases are to be regarded as truly magmatic, and 
not derived from surface-waters penetrating to the liquid lavas, as will 
be brought out later. They are here put forward as essential factors in 
the evolution of the magmas from the original planetary matter. 

The reactions working towards equilibrium are able to supply hydro- 
gen and carbon monoxide under conditions favorable to their absorption 
and retention, even if they were not originally present as occluded gases. 
The sources of the gases obtained from rocks are so complex that it is 
diflBcult to determine how much is to be assigned to each. Because of the 
penetration of surface-waters containing carbonic acid in solution, through- 
out the accessible rocks of the earth's exterior, it is likely that, in many 
eases, the bulk of the gas obtained by heating powders in vacuo has been 
<terived from acquired water and carbonated compounds. But in fresh 
meteorites, which presumably have not been subjected to action of this 
sort, occlusion is relatively more important. 

From the constitution of meteorites, some of the principles of early 
terrestrial evolution may, perhaps, be inferred, though the growth of the 


earth was probably not quite analogous, in all respects, to the formation 
of the meteorites. Whether we take the meteoritic material to repre- 
sent the heavier part of the original matter of the solar system, or the 
stellar system, as a whole, matters little in the geologic problem. If, in 
truth, the unoxidized, heterogeneously aggregated material of meteorites 
be typical of the original heavy material of the earth, it becomes evident 
that, in the case of our planet, other factors have been at work which are 
not operative in the bodies of which the meteorites are supposed to be 
fragments. These visitors from space are characterized by such minerals 
as cohenite, (Fe,Ni,Co),C, lawrencite, FeCl,, oldhamite, CaS,, and schreib- 
ersite, (Fe,Ni,Co)3P, which, next to nickel-iron, is the most widely distributed 
constituent of iron meteorites,^ though of less importance in the stony 
specimens. Such compounds imply an absence of both free oxygen and 
water in notable quantities. Of like import is the absence of hydrated 
minerals, such as micas and amphiboles. Water and an oxygenated atmo- 
sphere appear to be the agents which are lacking in the bodies from which 
the meteorites were derived, but which have been the operative factors in 
working over the outer portion of the earth. 

But the original source of the earth's atmosphere and hydrosphere 
is taken to be gas occluded, or absorbed, in the primitive meteoritic material. 
These original gases, escaping, furnished both atmosphere and hydrosphere 
when the earth became of sufficient size to retain them. A self-regulating 
system was inaugurated. In the early stages of the hydrosphere, when 
growth by infalling planetesimals was rapid, much water was buried 
within the fragmental crust. This material, worked over by volcanic 
activity, brought to the surface and subjected to weathering and erosion, 
and buried beneath more material, has undergone assortment and altera- 
tion until the accessible rocks at the present time are very different from 
the meteoritic matter. Since the earth attained its growth and the infall 
of planetesimals slackened, much less water has penetrated to great depths 
below the surface. Post-Archean sedimentaries have not yet reached 
thicknesses sufficient to carry inclosed water down to the depths from 
which the lavas arise. Deep mines indicate that fractures and fissures 
do not convey water down to very great depths at the present time. If 
water does not penetrate so rapidly now, and hydration and carbonation 
are less effective, it is also probably true that subsiding vulcanism brings 
less gas to the surface. 

It is essentially a system of balance. At the same time that water is 
being buried with sediment, its elements, hydrogen and oxygen, the latter 
in the form of the oxides of carbon, are exhaled from the earth's interior 
through volcanic outlets. But the system here suggested is very different 
from the postulated limited cycle of underground water which, following 
Daubr^'s famous experiment,' has crept into geologic literature as the 
origin of volcanic vapors and the modus operandi of vulcanism. Instead 
of surface-waters following cracks and fissures down to the hot lavas there 
to be absorbed, the water already is present, and is a part of the rocks and 

> Farrington, Jour, of GeoL, vol. 9, pp. 405-407 and 525-526. 

' Daubrte, ^Itudes Synth^tiques de G^ologie Exp^rimentale, 1. 1, pp. 236-246. 


magmas in the interior, whether actually combined as water, or as its 
elements held in solution, or chemically united in other compounds. These 
gaBeoua elements form an integral part in the magmas, having been vital 
dusters in their development from the primitive planetary matter. That 
this process of reworking has gone on to considerable depths, if we are to 
start with typical meteoritic material, is evidenced by the fact that the 
deep-seated plutonic rocks are characterized by micas and other hydrous 
minerals, while mineral species of the meteoritic type are absent.^ 

The more restrictive phase of the problem of water will be discussed 
under the head of vulcanism. 


In the actual dynamics of vulcanism, provided the gases are original 
in the magmas, the state in which they occur is not of vital importance, 
except in so far as it determines the conditions under which the gases be- 
come free, from occluded or chemical bonds, to perform their part in the 
mobility of lavas, in the explosions which sometimes accompany erup- 
tions, and in the phenomena of fumaroles and volcanic vents. The dis- 
tinction between cavity, occluded, and chemically united gas, which is 
made in the case of solid igneous rocks, can not be extended to the liquid 
lavas. In the liquid lava the gas may be supposed to be imprisoned 
mechanically, or else to form a part of the magmatic solution. On the 
solidification of the mass, the gas, formerly existing in the free state, may 
enter chemical cembinations at the lower temperature, may be occluded 
by the solid rock, or may become entrapped within the minerals last to 
crystallize. So, too, it is possible that some of the gas dissolved in the 
magma may, because of cooling and crystallization of adjacent portions 
of the solution, reach a supersaturated condition and appear in the solid 
rock also as gas inclusions. Otherwise, it would pass into the solid rock 
occluded or chemically combined. The condition of the gases examined 
in the laboratory need not, necessarily, correspond to a particular state 
of occurrence in the lava before crystallization. 

Gases mechanically distributed throughout the lava would always be 
an operative factor in vulcanism, while such gases as were chemically 
combined in the solution would, presumably, only become free, and hence 
fully operative, upon the lowering of the temperature and the relief of 
pressure,' and probably but partially then. Since vapors and gases in the 
free state are the cause of volcanic explosions, they can be traced as far. 
down in the conduits as explosions occur. From the nature of these explo- 
sions, which appear to be due to the accumulation of vapor gradually work- 
ing upward until suddenly able to relieve itself, it is fair to suppose that 

^This statement should perhaps be qualified. The basalt at Ovifak, Greenland, 
contains iron stronglv resembling the meteoric metal, and in which the minerals cohenite, 
lawrencite, and doubtfully schreibersite have been recognized. The occurrence of this 
terrestrial iron would indicate that material of this sort still occurs at points within the 
outer port of the earth. 

'A falling temperature favors the liberation of hvdrogen from water by ferrous 
compounds (see p. o7), while carbonates are most easily decomposed at low pressures 
(we p. 49). 


aqueoTis vapor and the auxiliary gases are present in the free state at 
still greater depths. 

It has been the observation of those who have studied volcanic erup- 
tions that water- vapor is by far the most abundant of the gaseous products 
of volcanoes. Water is also the principal compound of the element hydro- 
gen, which is quantitatively the most important gas obtained by heating 
igneous rocks in vacuo. According to one of the common theories of 
vulcanism, it is water, circulating underground and necessarily dissolving 
and absorbing mineral and gaseous material, which penetrates to the 
lavas and gives to them their supply of vapor and gases. Water, then, is 
a critical element in the theories of vulcanism, and likely to be a decisive 
factor, upon the basis of which many of these theories may stand or fall. 
It is, therefore, of great importance to know whether the aqueous vapor, 
which is so copiously exhaled from volcanic vents and plays such a rdle in 
vulcanism, is derived originally from the magmas, or is merely underground 
water which has been incorporated by the lava in its journey upward. A 
decision of this question will carry with it the solution of the allied question 
concerning the ultimate source of the other gases, and also throw much 
light upon some of the more comprehensive theories of vulcanism. 

Appealing to the fact that chlorine, in the form of hydrochloric acid 
and volatilized chlorides, is one of the products of volcanoes, one of the 
standard hypotheses attributes the cause of vulcanism to the penetration 
of sea-water to the heated interior. If this were so, isolated volcanoes 
far out at sea would be expected to yield much more chlorine than those 
on the continents. But the Hawaiian volcanoes exhale comparatively 
little chlorine or sublimed chlorides. It has been claimed that rain-water, 
sinking into the cone, would have sufficient head to exclude the sea-water 
from the neighborhood of the hot lava. Rain, however, falls upon but a 
small part of the whole cone, whose greater portion is under the sea. It 
would seem that if rain-water, falling upon a cone built up from the ocean 
bottom, is able, by means of its head, to keep out the sea-water which 
covers the lower slopes, the same amount of water precipitated upon a 
continental volcano would be even more efficient in preventing the general 
underground water from coming in contact with the lava in the conduit. 
Whatever may be the reason for the small amoimt of chlorine given off 
by the volcanoes of Hawaii, sea-water does not reach the heated lavas in 
sufficient quantities to affect them appreciably. 

On accoimt of the pressure exceeding the crushing strength of the 
rock, pores and crevices can not exist at depths greater than 30,000 feet 
according to the most generous estimate,^ and it is probable that continu- 
ous cracks cease much short of this. Beyond this extreme figure, meteoric 
waters can not be regarded as of any quantitative importance, on accoimt 
of the extreme slowness of diffusion through solid bodies not containing 
minute fractures. Liquid carbon dioxide still existing under great pres- 
sure in sand grains of Pre-Cambrian age is a concrete example of this 
slowness. While, theoretically, water may extend downward to the limit 
of the zone of fracture, the testimony of deep mining appears to show that 

^ HoBkios, 16th Ann. Kept. U. S. Geol. Surv., p. 853. 


meteorio waters grow relatively scant, as a rule, below the uppermost 
1,500 to 1,800 feet of the earth's crust.^ This shallowness of meteoric water 
increases the difficulties encountered by the hypothesis that the lava beds 
are supplied from this source, since they rise from far greater depths and 
only the upper portions of their conduits would be exposed to these waters. 

It is in this portion of the zone of fracture that Daubr^'s much quoted 
experiment upon the Strasbourg sandstone ' finds its application, if any- 
where, since numerous capillary pores with plenty of water are requisites 
for the operation of this principle. This famous experiment demonstrated 
that, owing to its force of capillarity, boiling water will pass through a 
disk of sandstone, 2 centimeters in thickness, against a slight steam-pres- 
sure on the other side. But it was only necessary for the steam-pressure 
to reach 686 millimeters, or nine-tenths of an atmosphere, in order to 
prevent any more water from passing through the sandstone. It is a long 
jump from this trivial capillary force, equal to less than one atmosphere 
of steam-pressure, to the great pressures which would have to be overcome 
in the depths of the earth's crust in order to reach the hot lavas, even though 
it be allowed that the water-vapor, if it came in contact with the lava, 
would be absorbed. Capillary force seems quantitatively inadequate. 

To reach the critical pressure of water due to the hydrostatic column, 
it is necessary to penetrate the earth to a depth of about 6,900 feet. At 
depths less than this, water passing into the vaporous condition, in the 
neighborhood of hot volcanic conduits, at temperatures below the critical 
point, should leave behind more or less of the matter held by it in solution, 
since the condensation, and hence molecular attraction of the vapor for 
solutes, is less than that of the water. Thus even if vapor from underground 
waters should enter the lavas, as Daubr^ has suggested, in the outer 
6,900 feet of the earth's crust, much of the chlorides, sulphates, carbonates, 
and silicates, dissolved in the water, would have been left behind. At 
depths between 6,900 feet and 25,000 feet, beyond which water can not 
penetrate, owing to the closure of all pores by the pressure of superin- 
cumbent rock, mineral matter dissolved in the water would probably still 
remain in solution when the liquid passed into the gaseous state at the 
eritieal temperature, since the density of the gas is equal to, or greater 
than, that of the liquid. 

The lava, being under considerable pressure, may be supposed to occupy 
aQ the cracks and crevices in the adjacent rocks, except those of capillary 
dimensions. If, therefore, in the passage of underground water into vapor, 
preparatory to entering lavas in the outer 6,900 feet of the earth's crust, 
much of the dissolved mineral matter be deposited in the minute pores 
leading to the lava, they should quickly become sealed, preventing any 
farther access, even of water, to the lava. To test this principle experi- 
mentally, a cylinder of medium-grained Potsdam sandstone from Wis- 
conon, 40 millimeters in diameter and 28 millimeters in thickness, was 
soldered into a short piece of iron piping, fitted at one end with an elbow 

* Kemp, Economic GeoL, vol. 2 (1907), p. 3; Finch, Proc. Col. Sci. Soc, vol. 7 (1904), 
pp. 193-252. 

' Daubrfe, Etudes Synth^tiques, t. 1, pp. 236-246. 


to serve as a receptacle for water, and at the other with a cork and a con- 
denser. When ready, the receptacle was filled with Lake Michigan water 
and a Bunsen burner was placed so as to heat the sandstone cylinder within 
the iron tube. One side of the sandstone was thus kept at a temperature 
slightly above 100°, while the other face, in contact with the water, remained 
just at the boiling-point. Water was found to penetrate the porous cylinder 
readily, evaporating and leaving its dissolved material within the mass 
of the sandstone, and escaping as steam on the farther side. The rate at 
which the water passed through the sandstone at the outset was not deter- 
mined, but after 5 liters of lake water had been used, it was found that 
129 cubic centimeters traversed the rock and were condensed in one hour. 
The rate slowly fell as the experiment progressed. While the thirteenth 
liter was being used, only 73 cubic centimeters passed through the sand- 
stone per hour. It was evident that the pores were becoming clogged, 
but to complete the experiment with Lake Michigan water, which contains 
only 150 parts of solid matter, per million, would have required too much 
time. To hasten the process, a saturated solution of calcium sulphate 
was substituted. This soon caused a marked slackening of the passage of 
water through the rock, and doubtless would have sealed the pores com- 
pletely, if allowed sufficient time. 

From thb experiment, it appears certain that water, evaporating in 
the pore spaces of a rock and escaping as steam, will leave behind what- 
ever material is in solution, until the crevices become clogged and the 
penetration of water ceases. This principle may be applied to the outer 
6,900 feet of the earth's crust; in the superficial portion of this zone it 
should be very effective, since the conditions more nearly approach those 
of the experiment; in the lower portion of this belt, as 6,900 feet and the 
critical pressure (as well as temperature in the neighborhood of hot volcanic 
pipes) is approached, the density, and hence the solvent powers, of the 
water-vapor approach those of the liquid. The vapor, also, should escape 
less readily from the liquid at these depths, since the expansive force of 
the vapor drives the water back along its path with more difficulty. Toward 
the critical point of water, therefore, the application of this principle 
becomes more uncertain, but it would seem to be operative also at these 
depths, though more and more slowly as the critical point is neared. 

It might be objected that the passage of water into vapor, involving 
the latent heat of steam, would keep the adjacent rocks cool and cause the 
deposition to take place at the very contact where the hot lava could fuse, 
and dissolve, the precipitated salts. But it is very doubtful whether the 
vaporization of such a small quantity of water, taking place with the slow- 
ness imposed upon it by the minuteness of the capillary pores, would keep 
the contact rocks at a temperature below 365°. The gap between 365° 
and 1100° is too great for there not to be a space, if of a few inches only, 
at an intermediate temperature. It is also to be remembered that the 
latent heat of steam diminishes with the pressure until, at the critical 
point, it becomes zero. The testimony of the country rocks through 
which a volcanic conduit has passed is that metamorphism has usually 
progressed to some distance from the contact of igneous intrusion. In a 
long-established volcano, where the rocks surrounding the conduit have 


been heated to high temperatures, the deposition of the solutes from any 
penetrating water should have sealed the capillary tubes and fissures at 
a distanee from the lava such that the latter cannot absorb them and 
keep the water-way open. Kemp has stated in a recent paper ^ that at the 
eontaets with eruptives, limestone rocks, instead of being porous, are 
prevailingly dense and compact, and often very hard to drill, as if due to 
deposition within their interstices. However, the author assigned this 
supposed deposition to magmatic waters from the intrusion. This brings 
up a widely established view that magmas, instead of absorbing water from 
the intruded rocks, give it off, depositing matter in solution to form veins in 
the sone of fracture. 
To quote Van Hise : ' 

In the bdt of cementation, in oonsequenoe of the porosity of that zone, the material 
of the magma, both by direct injection and by tranamisBion through water, may pro* 
foondly affect the aTerage chemical composition of the intruded rock for great distancee 
from the Intrusive mass. 

Geikie cites a case in Bohemia, where certain Senonian marls, invaded 
by a mass of Tertiary dolerite, begin to get darker in color and harder 
in texture at a distance of 800 meters from the contact, while, as the intru- 
sive mass is approached, the interstratified beds of sandstone have been 
indurated to the compactness of quartzite.* 

But considering only meteoric waters at depths greater than 6,900 
feet, where water remains liquid up to the critical temperature, it is less 
probable that the pore spaces will be filled up in this manner. Nor does 
it seem likely that Daubr^'s theory that water may penetrate rocks 
against a steam-pressure can operate at these depths, since that principle 
is dependent upon a marked difference between the capillarity of water 
and of steam, while at the critical point, the density of water-gas being 
the same as that of water, this force should be absent. The problem then 
becomes a question of equilibrium between the hydrostatic column of 
water and that of the lava, in which the pressure of the lava at a depth of 
7,000 feet should be in the neighborhood of 2.7 times that of the water, 
though this preponderance steadily diminishes as the water-gas becomes 
eondensed, with increasing depth, at a rate higher than lava. Whether 
under these conditions lava can absorb water-gas, is an open question. 

Water can only penetrate from 25,000 to 30,000 feet below the surface 
on account of the closure of all crevices by pressure. But on the assumption 
thai the temperature gradient in the outer part of the earth's crust is 1^ C. 
for each 100 feet of descent (which is probably too high) the critical tem- 
perature will not be reached, except in the neighborhood of volcanic intru- 
sions, until at a depth of about 36,000 feet. Hence, over the greater part 
off the earth, water will remain in the liquid state as far down as fractures 
and fissures will allow it to seep, and no appeal can be made to the more 
rapid and potent gaseous diffusion to carry it beyond 30,000 feet. But 
because of their heat, lavas must originate at much greater depths below 
the surface, and hence far beyond the reach of surface-waters, which can 

* Kemp. Economic GeoL, vol. 2, p. 11. 

* Van Hise, Monograph 47, U. S. G. S.. p. 714. 

'Hibsch, dted by Geikie, Textbook of Geology, vol 2, p. 774. 


only come in contact with them, and only doubtfully then, in a very limited 
portion of the throat of the volcano. 

These considerations seem to indicate that, for the most part, the volcanic 
gases and vapors have not been supplied to the lavas by ground waters, 
but are original constituents of the magmas. Doubtless at the beginning 
of an eruption, following a period of quiescence, much of the steam merely 
comes from such rain-water as may have accumulated in the crater and 
upper part of the cone, but this does not account for the gaseous emanations 
from the lava itself, nor from those volcanoes, such as Stromboli, and the 
well-known Solfatara near Naples, which maintain a mild form of eruption 
for long periods. Such meteoric water could contribute to the volcanic gases 
little except some dissolved air, together with a trace of carbon dioxide, 
and perhaps hydrogen from chemical action. Such soluble salts as this 
water might dissolve from the crater walls were brought up from the in- 
terior in the first place (making some allowance, however, for weathering), 
and so have little bearing on the case. 

The hypothesis that the gases and vapors are originally from the mag- 
mas, is greatly strengthened by the volcanic activity in the moon, if, as 
is rather generally believed, the great pits on the surface of the moon are 
craters produced by volcanic explosions; if not, of course the argument 
does not hold. The gases and vapors which caused the tremendous out- 
bursts can not be ascribed to the penetration of surface-waters and gases, 
for the moon has neither appreciable atmosphere nor hydrosphere, and, 
according to Stoney's doctrine, never could have held either, owing to its 
feeble gravitative control. Such gases as are implied by these explosions 
must be supposed to have arisen from within the interior of the moon. 
The extent of this explosive lunar vulcanism, in the absence of any appre- 
ciable atmosphere or hydrosphere, furnishes a strong argument against 
the belief that surface-waters and atmospheric gases are essential factors 
in terrestrial vulcanism. 

Thus far evidence of a negative nature has been brought forward to 
show the difficulties in the way of thinking that surface-waters play a 
prominent rdle in volcanic phenomena. But more positive evidence can 
be presented to support the view that the hydrogen and water in the deep- 
seated rocks are truly magmatic. Micas are prominent constituents of the 
plutonic rocks. The immense granitic bathyliths, which were probably 
formed beyond the reach of ground-waters, are characterized by this group 
of minerals. In fact, micas are more abundant in the deep-seated rocks 
than in the surface lavas of similar composition. Yet all micas contain 
hydrogen (or hydroxyl) and yield water upon ignition. This varies with 
the minerid species and locality, ranging up to 4 or 5 per cent. If these 
micas in the missive intrusions are primary minerals, as they seem to be, 
and were out of the reach of ground- waters until long after they were crys- 
tallized, there appears no other alternative than to consider this hydrogen 
as inherent in the magma itself. The general petrological principle that 
plutonic rocks are micaceous and hornblendic, while their more superficial 
equivalents are more frequently characterized by pyroxenes which are 
less hydrous, may point toward the suggestion that the magmas originally 
contain considerable water or the elements which can produce it, but as 


they approach the stirface much of the hydrogen and water-vapor escapes 
and pyroxene minerals crystallize instead of these hydrous micas. 

All of these facts and deductions lead to the general conclusion that 
our surface-waters have been derived from the interior of the earth, and 
oppose the idea that to explain the presence of hydrogen, or water, in 
magmas and rocks, we have merely to appeal to the penetration of surface- 
waters. The meteoric waters are limited to their superficial place and 
function, both in the evolution of magmas and in vulcanism; an ultimate 
source is found for these waters; and a steady supply of water and gases 
is furnished to the earth to offset the loss of vapor into space, and thus 
contributes to the globe one of the factors necessary to a long period of 
habitability for living organisms. 


The gases which escape from fumarolic vents are in many respects 
similar to those obtained by heating igneous rocks in vacuo, but with the 
addition of oxygen and vapors of chlorides, fluorides, boric acid, and other 
high-temperature volatilizations. Though nitrogen is much more con- 
spicuous in the analyses of volcanic gases than in those from rocks, this is 
doubtless due, in the main, to a mixture with atmospheric air. However, 
the greater heat of the volcano would also favor a higher proportion of 
nitrogen, as shown by my experiment. Much of the oxygen also is probably 
from the air. But an analysis of gas escaping from a stream of lava 
flowing on the sea bottom at Santorin gave Fouqu^: oxygen, 21.11 per 
cent; nitrogen, 21.90 per cent; and hydrogen, 56.70 per cent.^ This would 
suggest that the dissociation of water also contributes free oxygen. 

Fouqu^'s studies at Santorin confirm the law of variation in composi- 
tion of volcanic gases, first established by Sainte-Glaire Deville,' namely, 
that the nature of the gas evolved depends upon the phase of volcanic 
activity. Hydrochloric acid, with free chlorine and fluorine, is given off 
only from the hottest fumaroles where the heat is sufficient to liberate 
these gases from chlorides and fluorides. At less active vents, sulphur 
dioxide is the most noticeable of the corrosive gases, while the cooler fuma- 
roles exhale chiefly hydrogen sulphide, carbon dioxide, and nitrogen. 
Carbon dioxide and nitrogen escape from all the fumaroles. Fouqu6 
found that the relative importance of hydrogen increased with rise of 
temperature, and that his marsh-gas (which, owing to an imperfection in 
the method of analysis in 1867, may have been carbon monoxide, or a 
mixture of carbon monoxide and marsh-gas) diminished as the activity 
increased. These observations are entirely in accord with the results of 
my differential temperature experiments with rock powders. Hydrogen 
sulphide and carbon dioxide are the gases expelled from the rocks at the 
lowest temperatures; carbon monoxide and marsh-gas appear at inter- 
mediate temperatures, while hydrogen is most prominent when the heat 
is carried to bright redness. Nitrogen is most abundantly liberated at red 
heat; hence the presence of that gas at the cooler vents and fissures is 
chiefly due to atmospheric air. 

* Fouqu6, Santorin et see Irruptions, p. 230. 

> Sainte-Oaire Deville, Ann. de Chim. et Phys., 52 (1858), p. 60. 


While carbon dioxide escapes from all fumaroles in greater or less 
degree, it is at those vents whose activity has subsided beyond the point 
where hydrogen and the noxious gases are evolved that this gas is most 
conspicuous. For this reason, carbon dioxide has come to be regarded as 
marking the dying of the volcanic activity. A source for carbon dioxide 
after the disappearance of the other gases has been sought in the neigh- 
boring limestone formations, either from baking or from the chemical 
action of halogen or sulphur acids. The obvious difficulty confronting 
this conception is that limestone is not always present to furnish carbon 
dioxide. Experiments show that below 400^ G. carbon dioxide is the 
principal gas evolved from rock material, and as the lava solidifying in 
the crater, or conduit, has net lost all its gas, it is only a part of the natural 
sequence of events that the escape of carbonic anhydride from the cooling 
lavas should continue for some time after the volcano has settled into 
quiescence. Some of this carbon dioxide doubtless also comes from previ- 
ous lavas which, warmed again by the fresh lava, give up some of the carbon 
dioxide which my experiments show them to contain. 


Among the various substances which are deposited around fumaroles, 
sal-ammoniac, or ammonium chloride, is, in some respects, one of the most 
remarkable. Compounds of ammonium have not yet been recognized in 
igneous rocks, although rock powders often give o£f small quantities of 
ammonia gas when heated in vacuo. Chemical analyses of spring-waters 
report ammonium salts only in traces, such as may have been derived 
from the decay of organic matter. If ground-waters be, for the most part, 
unable to reach the lavas, even this rather doubtful source of ammonium 
compounds is not available. If the elements of the radical NH4 be supposed 
to have come from the interior magma, there are two alternative hypotheses 
still open. The first assumes that the radical NH4 existed intact in the 
magmatic solution in the form of ammonium salts and, volatilized by the 
heat upon the relief of pressure, gradually collected on the cooler portions 
of the crater. This hypothesis must, however, explain the apparent absence 
of these compounds in igneous rocks. The second believes that the am- 
monium chloride was formed synthetically in the throat of the volcano, 
from the nitrogen, hydrogen, and hydrochloric-acid gases. This would make 
it a direct product of volcanic gases. 

The presence of ammonia, or its vaporized salts, in volcanic emana- 
tions leads to the formation of another interesting compound. Silvestri ^ 
has foimd iron nitride, as a lustrous metallic deposit, at a fumarole on 
Etna. This compound is due either to a reaction between the sublimed 
ferric chloride and free ammonia gas or to the ignition of the iron-bearing 
lava in the presence of ammonium chloride vapor. The appearance of iron 
nitride around fumaroles throws no direct light upon the question of its 
existence in the magmas, though it indirectly leads to the hypothesis that 
the nitrogen in the ammonia and its compounds came originally from iron 
nitride within the magma. 

» Savertri, Pogg. Ann., vol. 167 (1876), pp. 166-172. 



The atmosphere id now being fed by gases which escape through out- 
lets other than those of active volcanoes. Work in the shafts of many 
deep mines in different parts of the world is often impeded by the exhala- 
tion of gases from the rocks. This iS; of course, familiar in the case of 
organic rocks, such as coal, in which the decomposition of organic substances 
is in progress. Reference is here made especially to gases escaping from 
crystalline or other inorganic rocks. An exhalation of this kind is a 
notable phenomenon in several of the mines in the Cripple Creek region of 
Colorado, where nitrogen and carbon dioxide are poured into the workings 
in considerable quantities when the barometer is low.^ Two analyses of the 
gas escaping into the Conundrum mine at Cripple Creek gave the following: 

1st: Carbon dioxide, 10.2; oxygen, 6.7; nitrogen, 84.1; total, 100. 

2d: Carbon dioxide, 8.3; oxygen, 10.2; nitrogen, 81.5; total, 100. 

No carbon monoxide, marsh-gas, or hydrocarbons were detected. 

The gas from the Elkton mine, which was analyzed by Dr. A. W. Browne, 
of Cornell University, consisted of nearly the same gases as from the Con- 
undrum mine: Water-vapor, 1.4; carbon dioxide, 14.7; oxygen, 5.6; nitro- 
gen, 76.8; argon, 1.5; total, 100.0. Hydrocarbons, methane, and hydrogen 
were absent.' The authors estimate that this gas may be considered to be 
25 per cent of air, 59 per cent of nitrogen and argon, 15 per cent of carbon 
dioxide, and 1 per cent of water-vapor. The gas apparently is derived 
from greater depths than those at which it issues, since it is warmer than the 
air of the mines, and since practically no gas was encountered in the oxidized 
zone. They regard the outpouring as the last exhalation of the extinct 
volcano, around whose neck the Cripple Creek mines are located. 

In some of the potash mines in the vicinity of Strassf urt trouble is caused 
by the escape of combustible gas into the workings. According to Precht,' 
blowers of this gas once lighted have burned continuously for periods as long 
as two months. An analysis of this gas by Precht shows it to be largely 
hydrogen. His figures are: Hydrogen, 93.05; methane, 0.778; carbon dioxide, 
0.180; carbon monoxide, trace; oxygen, 0.185; nitrogen, 5.804; total, 100.002. 
This investigator believed that but little of the hydrogen could have come 
from the decomposition of organic matter; instead, he sought a source for it 
in the possible oxidation of ferrous chloride in the salt by water, according 
to the reaction: 

eFea, + 3H,0 = 2Feaae + Fe,0, + SH, 

This source of hydrogen is somewhat analogous to the production of the 
same gas by the action of water upon ferrous compounds at high tempera- 
tures, which has already been discussed, except that in the salt beds the 
supposed action has taken place at the ordinary underground temperature. 
But these gases coming from the sedimentary salt beds of the Upper Per- 
mian represent, of course, gas merely restored to the atmosphere, and not 
an original contribution to it. 

^ Lindgren and Ransome, Prof. Paper 54, U. S. G. S., pp. 252-270. 
> H. Precht, Ber. Deutsch. Chem. GeeeU., vol. 12 (1879), pp. 557-561. 

* Loe, cit. 


Nitrogen with an abnormal amount of inert gas (probably both argon 
and helium) occurs, under high pressure, in a gas-well at Dexter, Kansas.^ 
However, instead of being derived from igneous rocks, this comes from a 
gas-bearing sand near the contact of the Permian with the Upper Carbonif- 
erous. An analysis of this gas gave:' Oxygen, 0.20; methane, 15.02; 
hydrogen, 0.80; nitrogen, 71.89; inert residue, 12.09; total, 100. 

Neither carbon dioxide nor carbon monoxide was present in this gas. 
The methane, and perhaps the hydrogen also, may be attributed to the 
decomposition of organic matter, since natural gas-wells exist at no great 
distance away. But the remarkable feature of this analysis is the large 
amount of nitrogen with the very abnormal percentage of inert gas. 
From this analysis, and the testimony of many spring-waters which give 
off considerable quantities of argon and helium, it would appear that 
gases often collect underground somewhat in proportion to their chemical 
inertness. The chemically active gases apparently are more largely retained 
within the rocks by combination, while nitrogen, having less power to 
unite chemically, more largely escapes from the rocks and accumulates in 
reservoirs. Argon, still more inert than nitrogen, thus may reach such a 
high proportion as 12 per cent. 


These studies show that, within the range of temperature employed, 
heat causes the expulsion of gases in whatever form they are held, and 
that the greater the degree of heat the more quickly and completely the 
gases are given off. There is reason to believe that this principle applies 
to the molten state as well as to the solid condition. If it be applicable 
to liquid lavas, it would favor the belief that a molten globe would have 
boiled out most of its gaseous matter before solidifying. Gases near the 
surface should escape rapidly. It might, perhaps, on first thought, be held 
that, while much of the gas in the outer portion would be lost, that exist- 
ing in the central part of the sphere would be retained and slowly recharge 
the peripheral portion after a crust had formed and prevented further 
escape; but the molten globe, by hypothesis, grew up gradually, and essen- 
tially every part was once superficial. Even to-day, in an essentially solid 
earth, there are movements of lava that bring up gases from unknown 
depths, and it is reasonable to suppose that the molten sphere was stirred 
up by still more effective convection currents which facilitated the expulsion 
of gases and vapors, and that almost all of the gaseous material of the 
globe would have been boiled out before solidification set in. 

The complete validity of this view depends much upon the fate of the 
gases after they have reached the surface. If they were retained in the 
form of a dense atmosphere, a condition of pressure-equilibrium might 
be established between the atmosphere and the gases in the liquid earth, 
by means of which the latter would retain some appreciable amount of 
gas. But if, as some believe, our atmosphere is about all that the earth 

1 Haworth and McFarland, Science, vol. 21 (1905), pp. 191-103. > Loc. cit. 


ean control/ the gas expelled from the molten sphere in excess of the mass 
of the present atmosphere would escape and be lost to the planet. Geo- 
logical evidences — early Cambrian glaciation, Paleozoic periods of aridity, 
and the general testimony of life — all point toward the conclusion that 
eariy terrestrial atmospheric conditions were not radically different from 
those of to-day. If the hypothesis of a heavy atmosphere be not permissi- 
ble! it becomes very difficult to explain the presence of original gases and 
gas-producing compounds in plutonic rocks on the basis of the Laplacian 
or other hypotheses that postulate original fluidity. 


After the gaseous matter of the ancestral sun was shot out from the 
solar surface to form the two arms of the spiral nebula, as postulated by 
the planetesimal hypothesis, the rock-producing portion is supposed either 
to have aggregated into planetesimal bodies, or to have been gathered, 
molecule by molecule, into the nucleus of the earth. The planetesimal 
bodies gathered in gas molecules of the atmospheric class both by chemical 
union and by surface adhesion or occlusion. As the earth grew by sweep- 
ing in the planetesimals, whatever gases they contained became entrapped 
in the body of the growing planet and well distributed throughout its 
mass. At first, the gravity of the earth may possibly have been able to 
hold only the gases brought in by planetesimal aggregates of rock material 
and those that became impounded in it by impact, but at a later stage, 
when increased mass enabled it to hold gaseous molecules, gases may have 
been added to the atmosphere directly from the nebula, and these, by 
chemical reactions, may have become united with the surface rocks. As 
soon as vulcanism commenced, a system of exchange was set up. While 
gases were being fed to the atmosphere by volcanic action, water, carbon 
dioxide, oxygen, and nitrogen were being buried with the surface rock 
material, partly by chemical union and partly by mechanical entrapment, 
as the growth by infalling matter continued. It is thus quite easy to 
understand how the earth came to be affected by these gases throughout 
its mass, and how they came to exist there in all available forms of retention. 

While the carbon monoxide and methane derived from rocks by heat- 
ing in vacuo are doubtless chiefly produced from the carbon dioxide and 
water present in the rock material, there seems good reason to suppose 
that similar reactions took place within the earth, as the surface material 
became buried and heated, and hence that carbon monoxide and methane 
exist, as such, in the earth's body, and are to be reckoned among the natural 
gases of the rocks. 


The fact that many of the igneous rocks are able to yield hydrogen 
firom reactions between water and ferrous compounds, at high tempera- 
tures, indicates that the material of the earth's crust is in a condition of 
partial oxidation only. Near the center of the earth there is probably 
very little oxygen, and even up to the surface, barring the weathered 

> R. H. McKee, Science, vol. 23 (1906), pp. 271-274. 


mantle, the rocks are suboxidized. Yet the earth is surrounded by an 
oxygenated atmosphere. Since oxygen is not developed in the combustion- 
tube, and does not appear to exist as a free gas in igneous rocks, it is not 
likely that this constituent of the atmosphere has come directly as an 
exudation from the interior of the globe. It is to be sought, rather, in a 
dissociation or decomposition of compound gases by physical or organic 
agencies. Originally, enough oxygen was derived from water-vapor, by 
physical means, to permit the beginning of plant life; after vegetation ap- 
peared, an abundant source of oxygen was found in the carbon dioxide. 

The average gas content of igneous rocks, as determined by the analyses 
now made, may be used to test the competence of the rocks to yield the 
present atmosphere. Taking the average volume of nitrogen per volume 
of rock to be 0.05, which b probably nearer the truth than the figure 0.09 
given in table 16 ^ (owing to leakage of air), it would require the liberation 
of all the nitrogen in the outermost 70 miles of the earth's crust to produce 
the nitrogen in the present atmosphere. For an estimate of the amoimt 
of igneous rock necessary to yield the carbon dioxide which is now locked 
up in limestone and coal deposits, we may take Dana's figure of 50 atmo- 
spheres of this gas, and an average of 2.16 volumes of carbon dioxide per 
volume of rock. To produce these 60 atmospheres of carbon dioxide, it 
is found that a thickness of 66 miles of crust would have to be deprived 
of its carbon dioxide' — a figure which corresponds fairly well with the 
estimate for nitrogen. If the water of the rocks be placed at 2.3 per cent, 
a depth of 70 miles would supply the hydrosphere. 

On the planetesimal h3rpothesis, gas has been supplied from the interior 
to the atmosphere ever since an early stage of the earth's growth, prob- 
ably from the earliest stage at which an atmosphere could be held, which 
' may be placed at the time when the earth's radius was about 2,000 miles. 
From this it appears that only a small fraction of the full gas-producing 
possibilities of the rocks of the earth was required to supply the atmo- 
sphere. The fact that gases are still being given forth through volcanoes, 
and that the ejected lavas still have gas-producing qualities, makes it clear 
that all the resources of the interior are not yet exhausted. The working 
qualities of the planetesimal hypothesis, therefore, do not seem to be found 
wanting in either past possibilities of supply, present output, or prospective 


In conclusion, I wish to express my special thanks to Dr. Julius Stieglitz 
for constant advice in the conduct of the chemical researches; to Dr. Oskar 
Eckstein for much valuable assistance in the laboratory; to Dr. R. A. 
IGllikan for helpful suggestions pertaining to physical principles and the 
designing of new pieces of apparatus ; and to my father. Dr. T. C. Cham- 
berlin, for proposing the investigation, and for constant sympathy and 
criticism during the progress of the work. 

' Ante, p. 28. ' The limestones, of oourae, are not here included. 





r TBI CAurKia ImrmmoM or Wuhukhov 





The Fobmeb Bates of the Eabth's Rotation and theib 
Bearings on its Deformation. By T. C. Chambesiin. 

The Botation-Pebiod of a Hetebogeneous Sphebold. 

By C. S. Sliohtbb. 

On the Loss of Enebgy by Fbiotion of the Tides. 

By W. D. MaoMillan. 

On Cebtain Belations among the Possible Changes in 
THE Motions of Mutually Attbacting Sphebes when 
Distubbed by Tidal Intebactions. By F. R. Moulton. 

Notes on the Possibility of Fission of a Contbacting 
Rotating Fluid Mass. By F. R. Moulton. 

The Beabing of Moleoulab Activity on Spontaneous 
Fission in Gaseous Spheboids. By T. C. Chambeblin. 


By Abthub C. Lunn. 


By Julius Stiegutz. 



Tbx FoBifBB Ratbb of thb Earth's Rotation and thbib Bbabingb on its 


The Astronomical DeductioDS d-20 

CoDsideratioDS Based on the Older Cosmogonies 6-8 

Considerations Based on the Planetesimal Theory 8-14 

Genesis of the Moon 14-20 

Evidences of a Present Change of Rotation 20 

Deductions from the Tides Theinielyes 21-46 

Tidal Phenomena of the Atanosphere 21-23 

Tides of the Uthosphere 23-25 

Pulsations of the Uthosphere 25-31 

Tides of the Hydrosphere 31-37 

More Radical Mode of Treatment 37-46 

Geological Eyidenoes 46-59 

Evidences from the Lithosphere 46-54 

Evidence from the Hydrosphere 54-57 

Minor Evidences 58 

Accelerative Agencies 58-59 

Conclusion 59 

Ths Rotation Pbriod of a Hstbroobnboxtb Sphbboid 63-67 

Thb Lobs of Enbbqt bt Fbiction of thb Tidbs 71-75 

Cbbtain Rblationb among thb Pobsiblb Chanobs in thb Motions of Mdtu- 

ALLT Atthacttino Sphbbbb whbn Disturbbd bt Tidal Intbbactions. 79-133 

Introduction 79-83 

General Ek^uations 84-86 

Lapladan Law of Density 87 

Moment of Loertia for the Laplacian Law of Density 88 

Special Case i-0 h-0 a,-0 <-0 5-0 89-94 

Application of Section V to the Ek^h-moon System 95-99 

Case i-0 h-0 i,-0 a,+0 «-0 5-0 100-104 

Application of Section VII to the Ek^h-moon System 105 

Application to Binary Star Systems 106-110 

Casei-O t\-0 i,-0 a,-0 «=»=0 5-0 111-115 

Application of Case X to the Earth-moon system 116 

Caset+O h-0 a,-0 «-0 5-0 117-118 

Api^cation of Case XII to the Earth-moon System 119 

Case i-0 i,-0 «-0 a,-0 5=1=0 120-125 

Secular Acceleration of the Moon's Mean Motion 125-126 

Summary 127-133 

NoTBS ON thb Pobbibiutt of Fission of a Contractino Rotating Fluid 

Mass 137-160 

Introduction 137-138 

Ellipsoidal Figures of Equilibrium of Rotating Homogeneous Fluids 139-141 

Poincar^'s Theorems Respecting Forms of Bifurcation and Exchange of Sta- 
bilities 142-144 

Figures of Equilibrium of Rotating Heterogeneous Fluids 145-147 

Ekjuations of Equilibrium for Constant Moment of Momentum 148-149 

Application to the Solar System 150-152 

Ai^cation to the Binary Stars 153-157 

Sunmuuy 158-160 



Bbabino of Molecular Activiti on Spontaneous Fission in Gaseous Sphe- 
roids 163-167 

Qbophtsical Theory under the Planstebimal Hypothesis 171-231 

SynopsiB 171 

Introduction 172-179 

Theory of Fisher .* 180-200 

General Equations 180-184 

History of the Compression 184-190 

Thermal Problem 190-200 

Critical and Supi^ementary 201-218 

Thermodynamic Theory 219-231 

The Relations of EQumnRixTM between the Carbon Dioxide of the Atmos- 
phere AND the Calcium Sxtlphate, Calcium Carbonate, and Calcium 

Bicarbonate of Water Solutions in Contact with it 235-264 

Ek^uilibrium in Aqueous Solutions of Calcium Carbonate, Calcium Bicarbon- 
ate, and Carbonic Add in Contact with an Atmosphere containing Carbon 
Dioxide, and the Solubility of Calcium Carbonate in Water containing Free 

Carbonic Acid 237-246 

The Determination of the Solubility Constant for Calcium Carbonate 245-249 

Equilibrium between Calcium Carbonate and Gjrpsum 250-255 

Calcium Sulphate, Carbonate, and Bicarbonate in the Presence of Sulphates. 255-258 
Calcium Sulphate, Carbonate, and Bicarbonate in the Presence of Sodium 

Chloride 259-262 

The Effect of Temperature Changes 262-263 

Summary of Results 263-264 





Ptofes»OT of Otology University of Chicago 


In the treatment of the earth's deformations, which is to be the subject 
of a following paper, it is essential to know whether changes in the rate of 
the earth's rotation must be regarded as one of the important factors or 
not. If the rate of rotation has appreciably varied during geological history, 
it is almost certain that the oblateness of the earth-spheroid has also 
varied, for unless the rigidity of the earth greatly exceeds that of any 
known substance, it must have been modified in form under changing 
rotation so as to approach the shape it would assume if it were a perfect 
fluid. It would be an error to assume, as is sometimes done, that the 
earth would conform to the fluidal shape perfectly, but that it would 
approach to this with a measurable degree of closeness seems to be beyond 
question. If there was a change from a high rotational speed, and con- 
sequent high degree of oblateness, to a slower speed with less oblateness, 
the surface area of the earth must have been reduced, because the nearer 
such a body approaches a sphere, the less the area of its surface, the greater 
its average gravity, and hence the greater its degree of compression. This 
is brought out numerically, with a high order of approximation, in the 
accompanjring paper of Professor Slichter. There will be occasion in the 
course of the present paper to consider in detail the application of this 
supposed reduction. 

Whatever therefore may be the difficulties attending a treatment of 
past rates of rotation of the earth, it is imperative that this element of 
the problem of deformation be recognized and evaluated so far as lies in 
our power. 

The problem may be approached on two rather distinct lines, one of 
which is astronomic but rests back so radically on postulates derived 
from theories of cosmogony that it may almost be called cosmogonic, and 
the other of which is geologic and rests on the direct or implied teachings 
of terrestrial e\adence. 

The ulterior purpose of this paper is to set forth the latter, but the 
cosmogonic considerations can not be passed without notice, for the cogency 
which will be thought to attach to geological evidences is certain to be 
measured in no small degree by the presumptions which are entertained 
on cosmogonic grounds, or on astronomic grounds with an essential cos- 
mogonic factor. The recent literature of the subject indicates that a 
belief in a former high rate of rotation of the earth based on cosmogonic 
and tidal grounds has a strong hold on astronomers and, to some large 



extent, upon geologists. The extent of this belief is due in large measure, 
no doubt, to the masterly papers of Sir George Darwin upon the origin 
and tidal influence of the moon. It is obvious that if the arguments in 
favor of a former high rate of rotation are accepted as decisive in them- 
selves, such geological data as seem to conflict with them are likely to be 
received with skepticism, or to be given interpretations consistent with 
the accepted conclusions. It is therefore appropriate, if not necessary, 
to review at the outset the grounds for the conclusions that have been 
drawn from cosmogonic postulates and from tidal and other considerations 
based upon these, so far at least as these have been thought to be weighty. 
There is the more reason for this in the present series of papers^ because 
of the very different basal postulates which may be grounded on the mode 
of planetary genesis set forth in them. 



It scarcely needs to be recited that, during the past century, astrono- 
mers and geologists almost universally accepted the hypothesis that the 
earth was formed from the condensation of a spheroid of gas, and that 
current doctrines as to the earth's early rates of rotation were founded 
on premises derived from some form of this hypothesis. Under the original 
Laplacian view it was affirmed that the rotations of the sun and the plan- 
etary masses were progressively accelerated as they shrank from a more 
expanded to a more dense condition. The rotation of the parent earth- 
moon spheroid was supposed to have reached, at a certain stage, such a 
velocity that a ring of matter was separated from its equatorial tract and 
formed the moon by subsequent condensation. It was held that the speed 
of rotation of the residual spheroid further increased, or tended to increase, 
by reason of its continued contraction, and hence that the primitive rota- 
tion of the earth was exceedingly rapid. As the present rotation of the 
earth is relatively slow, it followed, as a necessary inference, that a very 
marked decline in the earth's rotatory velocity took place in the course 
of geological history. 

In the modification of the Laplacian view introduced by Sir George 
Darwin,^ the material of the moon is supposed to have been separated 

^ G. H. Darwin: 

On the bodiljr tides of vibcoiib and semi-elafltic spheroids, and on the ocean tides 

upon a yielding nucleus. <Phil. Trans. Roy. soo. Lond., part 1, 1879, pp. 1-35. 
On tlu9 procession of a viscous spheroid, and on Uie remote history of the earth. 

<Phil. Trans. R07. Soc. Lond., pi^ 2, 1879, pp. 447-538. 
Fkoblems connected with the tides of a visoous splieroid. <Pliil. Trans. Roy. Soc. 

Lond., part 2, 1879, pp. 539-593. 
The determination of the secular efifects of tidal friction by a graphical method. 

<Proc. Roy. Soc. Lond., No. 197, 1879, pp. 168-181. 
On the secular changes in the elements of the orbit of a satellite revolving about a 

tidally-distorted planet. <Phil. Trans. Roy. Soc. Lond.. vol. 171, part 2, pp. 

713-891. 1880; Proc. Roy. Soc., vol. 29, 1879, p. 168, and vol. 30, 1880. p. 255. 
On the tidal friction of a planet attended by severad satellites and on Uie evolution of 

the solar system. <Phil. Trans. Roy. Soc. Lond., part 2, 1881, pp. 491-535. 
Enc. Brit., article on "tides"; ''The Tides." 1899. 
Also Thomson and Tait's Natural Philosopny, 2, articles on tides. 


from that of the earth after the condensation of the common mass had 
reached the liquid or perhaps even incipient solid state. In what precise 
form the separation took place is not specifically affirmed and is not material 
here, where the only essential point is the high rotatory velocity assigned 
the earth at the time of the moon's separation. 

Most or all of the meteoritic hypotheses of the earth's origin — using the 
term meteoritic in the restricted sense defined in this series of papers — 
agree essentially with the gaseous hypotheses in assigning to the earth, 
at its earliest separate stage, a molten condition and a rate of rotation 
either identical with or closely approximate to that of the Laplacian 
hypothec and of its modifications. The presumption, therefore, that the 
rotation of the primitive earth was of a high order of velocity had the 
sanction of these two classes of cosmogonic theories, and, as they occupied 
the field almost exclusively during the past century, this common inference 
from them came to have, naturally enough, a strong hold upon the beliefs 
of astronomers and geologists. If there shall finally be found reason to 
set these conceptions aside, it should still be recognized that they have 
been powerful instrumentdities in advancing knowledge and in stimu- 
lating inquiry, and that the investigations founded upon them have been 
scarcely less than necessary steps toward a final solution. 

Besides being at one in postulating a rapid rate of primitive rotation, 
these older hypotheses were essentially in agreement in assigning to the 
earth a molten condition in its early stages, as already stated, and this 
postulate has entered pervasively into the tidal and deformative theories 
of the earth that have had currency. Until the later decades of the last 
century, it was commonly believed that a molten condition was retained 
by the interior of the earth, or by some notable part of it, throughout 
the geological ages. In the latter part of the century, the conception of a 
solid earth came to be more generally entertained, but there went with 
this, almost universally, the postulate of such a degree of viscousness as to 
profoundly influence conclusions relative to tidal deformation and earth- 
movements generally. At the present time, when belief in an essentially 
solid earth has gained a large, though not universal, adherence, the con- 
ception that the spheroid is to be regarded as a viscous body in the treat- 
ment of all the larger geological problems is still widely prevalent and 
not only enters profoundly into the study of these problems but takes on 
forms exceedingly difiicult to adjudicate. The embarrassment does not 
arise so much from the theoretical recognition of a viscous property in 
the substances of the lithosphere, as from the lack of firm grounds for 
estimating its actual participation in the deformations and internal move- 
ments of the earth. One of the most vital questions of earth-dynamics 
relates to the respective values of viscousness and of elastic rigidity in 
terrestrial diastrophism. 

In this discussion the elastic rigidity of the earth will be regarded as 
the dominant factor in its morphology, and the tidal deformations of the 
lithosphere will be regarded merely as strains in an elastic body, involving 
viscous or liquid flowage only as an incident affecting those portions of 
the earth's body which are in a molten, gaseous, or temporarily unattached 


Working upon the cosmogonic grounds prevalent in the past century, 
and supported by the nearly universal consensus of opinion regarding the 
early stages of the earth, Sir George Darwin, in a memorable series of 
mathematical investigations/ developed the well-known doctrine of the 
tidal retardation of the earth's rotation from a primitive period of less 
than 6 hours 36 minutes to the present period of four times that length. 
Besides being grounded in presumptions that were commonly accepted, 
it had the merit of bringing these presumptions into historical consistency 
with the existing state of things. Not only that, but the investigation 
started with what then seemed to be a present rate of retardation deduced 
from astronomical observations, and proceeded backward by logical steps 
and current assumptions to the supposed original state, or at least to a 
close approach to it. The confidence that has been reposed in the conclu- 
sions so reached has not been placed without persuasive reasons, whatever 
conclusions may ultimately be reached from radically different cosmogonic 
postulates and from revised astronomical data. 

Not a few inferences of vital geological importance were drawn from 
this classic investigation, and specific data to support them were naturally 
sought in the geologic record. For the greater part, this search met with 
negative results, or with results which could be regarded as giving but 
meager or equivocal confirmation. Notwithstanding this, the logical force 
of the tidal argument as developed by Sir George Darwin, when its cos- 
mogonic postulates were taken for granted, was such that inharmonious 
geological phenomena were generally explained away, largely by assuming 
that the internal solidification of the earth took place at a relatively late 


As to the grounds for postulating a radically different constitution of 
the lithosphere, growing out of a new hypothesis of earth-genesis, I must 
content myself here with references to what has already been written' 
and to a fuller exposition elsewhere in this series of papers. It is appro- 
priate, however, to bring again to mind those inferences which are drawn 
from the rotational features now shown by the solar system, since these 
bear specifically upon the question in hand. 

The doctrine that a prevalent forward rotation of the planets could 
only mean that they were formed through gaseous or quasi-gaseous con- 
densation, was one of the bulwarks of the older hypotheses. It was only 

^ at., anU. 

> A group of hypotheses bearing on dimatic changes, T. C. Gbamberlin, Jour. QeoL, 

yol. 6, No. 7, 1897, pp. 6^-083. 
An attempt to test the nebular hypothesis by the relations of masses and momenta, 

Certain recent attempts to test the nebidar hypothesis, T. C. Cnamberlin and F. R. 

Moulton, Science, vol. 12. Aug. 10, 1900. 
The origin of the eartn, Chamoerlin and Salisbury, chap. I, vol. 2, Qeology, pp. 1-81 

Dec., 1906. 
Evolution of the solar system, F. R. Moulton, chap. XV, Introduction to Astronomy, 

pp. 440-487, liar. 24, 1906. 


when the inapplicability of this doctrine to natural cases was detected, 
about a decade ago/ that there was a clear path opened and logical grounds 
provided for developing a hypothesis of planetesimal accretion. It is 
pwhaps not too much to assume that the previous papers of this series 
have shown that this hypothesis is even better fitted than gaseous con- 
densation to give rise to the various rates of rotation actually presented 
by the solar system. Under the planetesimal hypothesis, the primitive 
rotation of the earth was not necessarily rapid, nor was the body of the 
earth necessarily molten. Thus two of the primitive conditions, that were 
formerly taken for granted on the basis of a nearly universal consensus of 
opinion, have been brought into question and may now be fairly regarded 
as being at best no more than working competitors with the alternative 
of a solid elastico-rigid earth, a view which is hampered by no compulsory 
presumption as to any particular rate of primitive rotation, but is hospitable 
to any rotational state which the direct evidences, astronomical, geological 
and otherwise, may require. 

The speculative freedom relative to primitive rotations, which the 
planetesimal hypothesis thus affords, directs attention anew to the actual 
facts and to their unembarrassed implications. The most fundamental 
case is that of the controlling body of the solar S3rstem itself. The present 
rotation of the sun is relatively slow and its axis is inclined appreciably 
to the common plane of the planetary system. When it is considered that 
the mass of the sun is more than 700 times that of all the planetary deriva- 
tives combined, this rate afld this inclination assume radical importance. 
This dow rotation and this inclination of axis are perfectly consistent with 
the planetesimal hypothesis and have peculiar suggestiveness in that 
relationship. On the other hand, they seem to me very difficult to reconcile 
with any theory under which the outlying bodies are supposed to be derived 
from a gaseous or quasi-gaseous spheroid by contraction, particularly 
any theory which postulates that the derived bodies were discharged from 
the central mass by the equatorial velocity of its rotation. Obviously 
the planetary material thus separated should be accurately adjusted to 
the sun's equatorial plane, and to the common plane of the system. Obvi- 
ously also the great residual mass should have a rate of rotation appro- 
priate to such a discharge. Having separated a succession of masses 
from its equator to form the planets, and having further shortened its 
radius some 36,000,000 miles after the last known planetary mass was 
detached, the sun should have a rotatory velocity somewhat near that 
requisite for another planetary separation. The velocity of rotation at 
the equator of the solar nebula when it was supposed to have detached 
the material for Mercury must have been, according to the Laplacian hy- 
pothesis, about 28 miles per second. The equatorial velocity requisite to 
bring the centrifugal and centripetal components of the sun's equatorial 
motion into equality if the sun now had a radius of 1,000,000 miles is 176 
miles per second; the velocity required to bring about this state at the 
present surface of the sun is 270 miles per second. We should then expect, 

^ Journal of Geology, vol. 5, 1897, pp. 009-609 


under any hypothesis that rests on centrifugal separation, that the present 
speed of the sun's equator would certainly be much greater than 38 miles 
per second, and should approach the higher figures given. As a matter of 
fact, the Sim's equatorial velocity of rotation is only about 1.3 miles per 
second. Such a rate seems, therefore, to be altogether inconsistent with 
the doctrine of centrifugal separation. If, for a moment, the thought be 
entertained that tidal retardation may have reduced the sun's rotation 
from a high primitive rate consistent with the centrifugal hypothesis, to 
the present rate, it will become obvious, on a study of the nature and value 
of the tidal influence of the planets on the sun, that this is wholly unten- 
able. The quantitative estimates of Sir George Darwin are decisive on this 
point.^ So also is the remarkable fact that the equatorial portion of the 
Sim has a higher rotational velocity than the portions in higher latitudes 
instead of lagging, as it should if it were affected by tidal retardation. 

The obliquity of the sun's axis is a further grave objection to all forms 
of the doctrine of centrifugal separation. On the other hand, some such 
obliquity is extremely probable under the hypothesis that the system was 
developed by the influence of a passing star, for the axis of the ancestral 
sun might obviously sustain any relationship to the orbital plane of the 
disturbing body. The position of the present axis, under this hypothesis, is 
the result of a composition of moments of momentum derived in part from 
the ancestral rotation and in part from the passing star, and it could not 
therefore be expected, except by a remote chance, to be exactly normal 
to the common plane of the planetary system. Under this h3rpothesis the 
obliquity of the sun's axis, together with its slow rotation, suggest, if they 
do not distinctly imply, that the direction of rotation of the ancestral sun 
was opposite to that of the present sun, and that its axis was more inclined 
than now to the plane of the present system. 

Of similar rotational import is the relationship between the time of 
rotation of Mars and that of the revolution of its inner satellite, Phobos. 
It is obvious that, under any hypothesis of centrifugal separation, if a 
revolving spheroid acquires by contraction an equatorial velocity sufficient 
to leave behind the material of a satellite, and afterwards continues to 
contract until its radius is but a small fraction of its value at the time of 
separation, the rate of rotation of the spheroid must be greatly increased 
and its period must be much shorter than the revolutionary period of the 
derived satellite, unless some very potent agency intervenes to reverse 
the systematic process of the evolution. Now, the satellite Phobos revolves 
around Mars about three times while the planet rotates once. In an anal- 
ogous way the little bodies that make up the inner edge of the inner ring 
of Saturn revolve about that planet twice while the planet rotates once. 
These are, on their face at least, seriously out of accord with the doctrine 
of centrifugal separation by planetary contraction. Darwin has suggested 
that tidal retardation may be a possible solution in the special case of 
Phobos, but Moulton has called attention to the insuperable difficulties 
of applying this explanation consistently to the Satumian case and the 
Martian case at the same time.' 

» Trans. Phil. Soc. Lond., 1881. * Astrophya. Jour., vol. 11, Mar., 1900, p. 109. 


The retrograde revolution of the ninth satellite of Saturn still further 
and seriously complicates the case from the centrifugal point of view. 

The rotations of Uranus and Neptune are unknown, but, whether they 
are concordant with the revolutions of their satellites or not, they present 
difficulties under the centrifugal hypotheses because of the great obliquity 
of the planes of revolution of their satellites to the common plane of the 
sjrstem, and because of their retrograde motions. This has long been 
recognized, but these difficulties gain not a little in force when they are 
associated with other rotational difficulties which have been insufficiently 
considered in connection with them. 

There are still other rotational features of the existing planets which 
seem to be inconsistent with all forms of a contractional-centrifugal hypoth- 
esis of planetary origin, and, what is especially to the point in the matter 
in hand, which seem inconsistent with the very high rotational velocities 
which such a hypothesis necessarily postulates. Among these are the great 
differences in the rotational features of the members of the system. If 
the system started from a common spheroid and if its derivatives were 
shed by systematic centrifugal action, it is very difficult to see how so great 
variety of rotational velocities, so varied inclinations of the rotational 
axes, and so diverse directions of rotation as the system actually presents, 
could have arisen. 

Under the planetesimal hypothesis, the rotation of each planet is held 
to have arisen independently of every other planet. Its rate of rotation 
depended on the special conditions that attended the expulsion of its 
nucleus from the sun, and on the mode of accession of the rest of its material 
from the planetesimal state — conditions that were quite certain to vary 
with each planet. Rotations, rapid or slow, direct or retrograde, with 
inclinations of any degree, are consistent with the hypothesis. There is, 
however, a decided balance of presumption in favor of forward rotations, 
of moderate inclinations of axis, and of moderate velocities of rotation. 
As none of the planets rotate at speeds that even remotely approach that 
requisite for equatorial discharge, as their rates of rotation differ widely 
frbm one another^ as the inclinations of their axes vary greatly, and as the 
majority of their rotations are direct and the minority retrograde, this 
hypothesis seems to be concordant with the facts of the case. 

If we are thus permitted to start with a genesis which leaves us free 
to suppose that the rate of the earth's rotation at the outset may have 
been essentially what it is to-day, or may have been faster or slower in any 
degree, the preconceptions that have led to former rotational views do not 
trammel us. The determination of the past history of the earth's rotation 
rests unhampered upon the evidences presented by its own phenomena 
and upon those deducible from the necessary influence of its neighbors. 

The most important of the rotatory influences of neighboring bodies on 
the earth is the friction of the tides, particularly of the lunar tides. The 
assigned mode of this action is familiar and may be stated briefly as fol- 
lows: If the tidal protuberance has a position in advance of the position 
of the moon, as at A in fig. 1, a component of the moon's attraction tends 
to antagonize the earth's rotation and to accelerate the moon's motion. 



If a tidal protuberance has a position behind the moon's position, as at 
jB, a component of the moon's attraction tends to accelerate the earth's 
rotation and to retard the moon's motion. If a protuberance rises directly 
beneath the moon's position, its forward and backward pulls are equal. 

On the opposite side of the moon 
there are complementary protu- 
berances, as A'^ B\ fig. 1, whose 
rotatory effects are the reverse 
of those on the moonward side, 
but whose greater distances give 
them less efficiency. It is merely 
this difference in effectiveness 
growing out of difference of dis- 
tance that is usually appealed to 
as influencing rotation. 

I shall endeavor to show later 
that, while the foregoing reason- 
ing seems to be unimpeachable in 
itself, there are counterbalancing 
factors which seem to have been 
overlooked, and which nullify the 
value of this mode of treatment. 
They do not, however, nullify 
the proposition that tidal friction 
tends to retard the earth's rota- 
tion. It seems best, however, to 
review the subject first on the 
accepted lines. 

The tides represented in fig. 1 
are such as are assigned to the 
direct pull of the tide-producing 
body and are known as ''direct.^' 
The protuberanpe A, fig. 1, repi^ 
sents a tide which is int^^reted 
as having lagged in its forma- 
tion, and hence has been carried 
forward by the rotation of the 
earth to a position in front of 
the moon's position; B represents 
a tide which has been formed 
behind the moon's position, but 
both may be regarded as falling within the class of '^ direct" tides. This 
class of tides are said to be built up when the natural free period of the 
tidal wave is less than that of the tidal forces. If the natural free period 
of the tidal wave is greater than that of the tidal forces, the tendency is to 
produce '' inverted" tides. The law underlying this difference of result, to 
which Newton first directed attention, is thus stated by Darwin:^ 

Fio. 1. 

» The Tides, pp. 171-172. 



Now, this ample esse illustrfttos a general dynamioal principle, namely, tbat if a 
qrttom capable of oeoiUating with a certain period ia acted on by a periodic force, when 
the period of the force is greater than the natural free period of the system, the osdl- 
laiions of the system agree with the oscillations of the force; but if the period of the 
loroe is less than the natural free period of the system the oscillations are inverted 
with reference to the force. 

This principle may be ap- 
plied to the case of the tides in 
the canal. When the canal is 
more than 13} miles deep, the 
period of the sun's disturbing 
force is 12 hours and is greater 
than the natural free period of 
the oscillation, because a free 
wave would go more than half 
round the earth in 12 houra. 
We conclude, then, that when 
the tide-generating forces are 
trying to make it high water, it 
wfll be high water. It has been 
shown that these forces are tend- 
ing to make high water immedi- 
ately under the sun and at its 
antipodes, and there acccmiin^y 
win the Idgh water be. In this 
case the tide is said to be direct. 

But when the canal is less 
than 13} miles deep, the sun's 
disturbing force has, as before, 
a period of 12 hours, but the 
period of the free wave is more 
than 12 hours, because a free 
wave would take more than 12 
hours to get half round the 
earth. Thus the general prin- 
ciple shows that wtoe the forces 
are trying to make high water, 
there will be low water, and vice A 
versa. Here, then, thm wOl be 
low water under the sun and at 
its antipodes, and such a tide 
is said to be inverted, because 
the oscillation is the exact in- 
version of what would be natu- 
rally eiqiected. 

All the oceans on the earth 
are very much shallower than 
fourteen miles, and so, at least 
near the equator, the tides ought 
to be inverted. The conclusion of 
the equilibrium theory will therefore be the exact opposite of the truth, near the equator. 

This argument as to the solar tide requires but little alteration to make it applicable 
to the lunar tide. 

The positions of a set of ''inverted" tides corresponding to the fore- 
going set of "direct" tides are shown in fig. 2. 

Now since the rotation of the earth gives its surface an angular motion 
greater than that of the tide-producing body, its effect must be to carry 

Fzo. 2. 


the protuberances forward towards positions which, according to the 
above interpretations, are in some cases more favorable for influence on 
the earth's rotation and in other cases less favorable, and if a wave is followed 
through its whole course, it sustains various relations, favorable and unfa- 
vorable, retardative and accelerative. The sum total of influences is thus 
seen to be a product of much complexity. It has been held that the retar- 
dative positions predominate in effectiveness. The case has usually been 
treated on the assumption of a continuous ocean belting the earth and 
permitting the tides to follow the tidal forces consecutively about the 
earth. It will be shown later that this is not the actual case, and that 
the tides are essentially limited to individual water-bodies. This further 
and greatly complicates the case. The problem is still further complicated 
by the past relations of the moon to the earth, and this claims attention 
before further considering the eflSciency of the tides. 


(1) The hypothesis of Laplace in its original form took no account of 
tidal action. Under it the rotation of the earth, when it had become 
condensed to a molten globe, was assumed to have had the velocity which 
at .an earlier stage was necessary to separate the lunar ring, plus that 
which was added by subsequent contraction. How this high rate of rota- 
tion was reduced to the existing rate was not explained. 

(2) The supplementary hypothesis of Sir George Darwin replaces this 
defect of the Laplacian hypothesis by postulating a centrifugal separation 
of the moon-mass from the earth-mass after the parent-body had been 
condensed to a liquid or perhaps even incipient solid state, and a subse- 
quent recession of the moon by tidal influence, accompanied by a reduction 
of the earth's rotation as its dynamic reciprocal. The postulated method 
of this tidal action has been stated above. The fundamental proposition 
that tidal friction will tend either to separate the two interacting bodies 
or to draw them together — according to the precise nature of their rela- 
tions — is not questioned, as it seems to be solidly founded on the laws of 
energy, but it is necessary to consider the precise relations of the bodies to 
determine the character of the action under the preceding mode of inter- 
pretation, and we shall find occasion to question the mode itself. Darwin's 
method of starting with what was thought to be a fairly reliable astronom- 
ical indication of the present value of the earth-moon interaction and of 
working backward mathematically to the primitive state, or so far as the 
mathematical process would carry, is beyond praise. But as the present 
value of the earth-moon interaction is open to serious question and is not 
now replaceable by an unquestionable value, and as the postulates for the 
backward tracing are themselves in question, it is necessary to consider 
the hypothesis on more general lines. The value assignable to the tides in 
each of the earth's ages depends on the assumptions made regarding the 
physical states of the earth's interior. If the body of the earth be assumed 
to be molten, or viscous in such a degree that the body tides are important 
and are of the liquid or viscous type, the results will be very different from 


those which will be reached on the assumption that the body tides are 
merely strains in an elastico-rigid earth. So, too, if the ocean has been 
growing in volume during the geologic ages, or has been changing in form 
in any notable degree, the results would need to be modified accordingly. 
EiVen when approached on the admirable lines of backward tracing by 
computation, the results are therefore subject to wide variation accord- 
ing as the postulates arising from one cosmogonic hypothesis are used or 
those of another. 

If we were to follow carefully the first stages of the moon's evolution 
under Darwin's hypothesis, it would be seen how critically dependent 
that hypothesis is on an underljring theory of cosmogony. The moon is 
assumed to have been separated from the parent earth-moon mass by 
some form of centrifugal action. While the precise form may have been 
either one or another of two or more alternatives, the principle of action 
is the same up to a certain point and is best illustrated by supposing that 
the moon-mass separated as a unit, and that just after separation it was 
a spheroid close beside the earth-spheroid and revolving in the period of 
the latter's rotation. An objection to this supposition •will be considered 

Now at this critical stage the earth was subject to the tidal action of 
the sun, which, according to the fundamental theory of the hypothesis, 
should tend to retard the earth's rotation. The earth was also subject to 
contraction from loss of heat, which should tend to accelerate its rotation. 
If the former was the greater influence, the lunar tide, which would have 
begun to be generated as soon as a difference arose between the moon's 
revolution and the earth's rotation, must have fallen behind the moon's 
position and, according to the hypothesis, must have tended to draw it 
backward and bring it down to the earth. To permit the evolution to 
proceed at all it m necessary to suppose that the contraction from loss of heat 
was a greater influence on the earth's rotation than were the solar tides. Now, 
the earth's contraction from loss of heat at the present time is exceedingly 
small. If, therefore, the constitution of the earth has been much the 
same as it is now ever since its growth practically ceased, as assumed by 
one cosmogonic hypothesis, the supposition that the tidal evolution of the 
earth-moon system was started in the right direction for lunar recession 
by the superiority of the influence of contraction over that of the solar 
tides is either untenable or else the solar tidal influence was extremely 
small. The initiation of the tidal evolution postulated by this hjrpothesis 
is thus seen to be tied up with a very high rate of loss of heat in the initial 
earth-stages, and this is only assignable under certain cosmogonic assump- 
tions which give to the earth a very hot surface. These are, however, 
not necessarily confined to the gaseous or meteoritic hypotheses. They 
may possibly be made under the planetesimal hjrpothesis, but in any case 
they are as speculative as the hypotheses themselves, and in the latter 
case somewhat less well grounded, because the alternative phases of the 
hypotheses seem to be the more probable. 

It is worth while to note further in this connection that the evolution 
of the lunar tide, under the theory of Darwin, would be a very slow process, 


and might be reversed before it escaped the critical conditions named 
above. The speed of the earth's rotation and the speed of the lunar tide in 
this early stage must not be conf ounded, for the speed of the tide is depend- 
ent on the difference between the angular rate of the earth's rotation and 
the angular rate of the moon's revolution. At the instant of separation 
the two rates were the same, and the tide, if the stationary protuberances 
caused by the moon can be called a tide at all, would be infinitely slow, 
and the period of the tide — ^the time required for the tide to make a circuit 
of the earth — ^infinitely great. At the first stage of difference in angular 
rate, if the earth's rotation was accelerated by contraction more than the 
solar tides retarded it, the movement of the tide would be infinitesimally 
slow in the present direction, and the tidal period sub-infinitely long. The 
movement of the tide over the face of the earth would be accelerated only 
as the contractional acceleration continued to be superior to the solar 
tidal retardation reinforced by the lunar retardation. If the rate of loss 
of heat, which must have declined rapidly as the supposed molten earth 
crusted over and the crust became thicker, fell below that at which its 
accelerating value on the rotation of the earth was superior to the retar- 
dational value of the solar and lunar tides, and the latter then became supe- 
rior, the time of rotation of the earth might be forced back to coincidence 
with the revolution of the moon, and the lunar tides temporarily suspended 
and, a little later, reversed, and the moon brought back to the earth accord- 
ing to the fundamental postulate of the theory.^ To escape this contin* 
gency it is necessary to suppose that the contractional influence of the loss 
of heat continued to be superior to the retardational influence of the solar 
and lunar tides until the lunar tide, though developed with extreme slow- 
ness, had extended the moon's revolutionary period so much that when 
the retardational influence became superior to the contractional influence 
it was too late for it to reduce the rotation-period of the earth to the revo- 
lution-period of the moon at that end of the evolutional series. These 
considerations serve to indicate how delicately poised were the initial condi- 
tions assumed by the hypothesis and how completely they were dependent 
on the heat-emission of the earth at the critical stage, which in turn was 
dependent on the cosmogony that preceded it. 

The argument that the balance of influence must have lain on the side 
of heat-loss or else the moon would not be where it is at present, would 
be pertinent if the earth-moon evolution were absolutely shut up to one or 
the other of the alternatives just considered, but it has no force against a 
hypothesis which entirely avoids these critical alternatives. 

It is to be noted further that the tidal reactions in the initial stages of 
the hypothesis of Darwin must apparently have been those of the earth's 
body, for if the heat had been so far dissipated that the earth was crusted 
over and the oceans were permitted to mantle the earth, the loss of heat 
would possibly have been too small to start the evolution in the postulated 
direction. The atmosphere must then have been of that vast vaporous 

* The correctness of this is dependent on the soundness of the theory that the vonUon 
of the tides determines their acoelerative or retardative character, which will be con- 
sidered later. 


kind made so familiar to us by the geologic rhetoric of the last century. 
Until the outward reach of this atmosphere was escaped by the receding 
moon, the atmospheric friction must have kept the moon in consonant 
revolution with itself and tidal action could not have been inaugurated. 
This must have prolonged the critical stages and made the triumph of 
contraction over the solar tides all the more doubtful. 

The separation of the moon from the earth after the common mass 
became a liquid spheroid is subject to another serious contingency, based 
upon the same principle ol differential attraction as the tides themselves. 
Roche has shown that a satellite revolving within a given distance from its 
primary will be torn to fragments. The fragments must revolve at velocities 
strictly dependent on their distances from the center of the primary and 
hence must disperse themselves into a ring of the Satumian type.^ The 
fragments so produced would be subject to further reduction by collisions 
with one another, by changes of temperature, and by internal reactions, 
and would probably only reach an approximately stable condition as to 
size when they were well comminuted. For the earth-moon combination 
the Roche limit of disruption lies about 11,000 miles from the earth's 
center.' The cogency of Roche's reasoning, supported by that of Clerk- 
Maxwell and others, and the example of the rings of Saturn, seems to leave 
no alternative but to suppose that a body of the mass of the moon could 
not pass from the earth outward by tidal reaction without being torn to 
fragments and converted into a ring, unless the fission of the earth-moon 
mass and the initiation of the lunar tide took place outside the Roche 
limit, which is difficult to believe under the Darwinian hypothesis, though 
consistent enough with the Laplacian. The laws of revolution seem to 
forbid the supposition that the fragments produced by tidal disruption 
could have been aggregated for any appreciable length of time on one 
side of the earth so as to act jointly in producing an effective tide. Even 
if some tide could be so produced there would still remain the question 
whether it would have carried the fragments outward by reaction suffi- 
ciently far for them to have escaped the dangers of reversal by the solar 
tides, as pointed out in a preceding paragraph. There seem to be no 
cogent theoretical grounds upon which it can be affirmed that the frag- 
ments of a disrupted body of this kind would evolve into any other condi- 
tion than that of a ring of discrete particles during the time available for 
starting the recessional movement of the moon. They might perhaps 
move outside the Roche limit or be drawn down to the planet in a period 
sufficiently long, but probably not in the available period. The fact that 
the Satumian rings are present at this stage in the history of the solar 
system suggests, if it does not definitely imply, that this form of organiza- 
tion is one of much persistence. 

If we pass by these peculiar difficulties that embarrass the supposed 
separation of the moon from the earth, and if we set aside the special 
consequences assigned to a molten or viscous earth-body, the remaining 

^ On the stability of motion of Saturn's rings. < Scientific Papers of James Cleik- 

:wdl, vol. 1, pp. &8-376. 

* Darwin's ^' Tides," pp. 368-360. 


problems respecting the influence of the tides on the earth's rotation are 
essentially the same, whatever the genesis of the moon, and so these further 
problems may well be deferred until it is seen where and how the various 
genetic theories come onto common ground and may be treated in a com- 
mon manner. 

(3) There is nothing in the planetesimal hypothesis that is, in itself, 
necessarily prohibitive of an origin of the moon by centrifugal separation 
from the earth-mass, for under it the planetary bodies may have had very 
high rates of rotation. So also the mass of the planetary nucleus may 
have been so large and the ingathering of the planetesimals may have 
been so rapid, by hypothesis, that a molten or even a gaseous condition 
could have arisen. In the case of the larger planets such a primitive state 
is quite within the limits of the probabilities. The case of the earth is 
debatable, but it will be of no service in this discussion to follow the 
gaseous or molten alternative, as it would be essentially identical with the 

There are two other possible modes of origin of the moon, in neither of 
which was the moon-mass ever a part of the earth-mass. In both of these 
it is supposed that the nebular nuclei of the earth and the moon were 
separate knots of the parent spiral nebula. In the first case, they are 
supposed to have been companions in projection from the ancestral sun, 
and to have revolved about their common center of inertia from the out- 
set. In the second case, the nuclei are supposed to have been at the outset 
independent knots having separate orbits about the sun but near one 
another. The two are supposed to have come into their present relations 
in the course of the segregation of the parent nebula. Rather grave 
dynamic difficulties attend this latter view, and it need not be pursued 
further here, as the rotatory problems under it are not essentially different 
from those of the first and much more probable alternative. 

In this preferred alternative, the nuclei of the earth and moon, at the 
instant they left the ancestral sun, are supposed to have been a single mass 
which was given a forward rotation by the unequal resistances on its oppo- 
site sides to the expelling impulse, for which there are assignable reasons. 
Just after leaving the sun, the mass is supposed to have separated as an 
incident of the expulsion, but the two parts are supposed to have continued 
to revolve about their common center of inertia essentially as before, i.e., as 
a rigid body. After separation, however, each was subject to the rotational 
effects of the accession of planetesimals, and when their rotations came to 
differ from their revolution about the mutual center of inertia they were 
subject to tidal reaction. The extent to which such differences of rotation 
arose is an essential part of the problem under this hypothesis. 

There were many possible alternatives, theoretically speaking, as to 
the relative sizes of these nuclei and the distances to which they separated 
under the initial impulse, but there were limitations to these. If the com- 
bined masses of the two nuclei were one-eighth of the combined mass of 
the present earth and moon, the moon could not have been more than 
460,000 miles from the earth, but as this is farther than it is at present the 


hypothesis is not hampered by this limitation. If the joint mass was 
larger, their initial distances may have been greater; if smaller, they 
must have been less. Within the limits thus imposed by the mechanics 
of the case, the nuclei may have been separated by any distance, abstractly 
considered, from the maximum permitted down to siu^ace contact. 

In fact, however, the degree of nearness consistent with the present 
state of things, was limited by the consequences of growth, for the increase 
of the masses of the nuclei by the ingathering of the planetesimals may 
have drawn the nuclei toward one another, or even together. This was 
conditioned by the moment of momentum which the accessions carried 
into the nuclei, which varied widely. Separation later by centrifugal 
action would be theoretically possible, under assignable conditions, but not 
at all inevitable, perhaps not at all probable. If the initial distance of the 
nuclei were sufficient, however, the nuclei might approach one another so 
long as growth was a ruling influence. Tidal action would nm concur- 
rently with this and would oppose approach, under most conditions, but 
during the more rapid stages of growth, the tidal effect may possibly have 
been less than the effect of increasing mass. But the tidal effect would 
increase as the bodies were drawn toward one another, while in the later 
stages of growth the increase of the mass would decline in rate. At a certain 
stage the two effects may be presumed to have balanced one another, after 
which recession would begin through the preponderance of the frictional 
effect of the tides. From that stage, the history would proceed along the 
lines determined by the mutual interaction of the matured bodies.^ 

It will be seen that the range of specific assignments under this phase 
of the planetesimal hypothesis has a wide amplitude, embracing the per- 
missible assignments as to the original distances between the nuclei, as 
to the original masses of the nuclei, and hence as to the amount of their 
growth, as to their planes of revolution, their eccentricities of orbit, etc. 
Under this amplitude, it is possible to suppose that the two bodies at the 
climax of their approach reached precisely the relations which were indi- 
cated by Darwin in his backward tracing of their history. On the other 
hand, so far as the h3rpothesis itself is concerned, it is equally possible 
that the approach of the bodies was much less close, and hence that their 
recession under tidal influence was correspondingly less. It will be seen, 
therefore, that this h3rpothesis has very much greater adaptability than 
the hypothesis of centrifugal separation, and does not equally hamper us 
respecting subordinate hypotheses, such as a molten state, a viscous 
interior, or a particular amount or a particular distribution of the hydro- 
sphere. We are quite free to follow backward from the present observa- 
tional data, when it shall be possible to do this on firm ground, with the 
utmost complacency as to the results, and to accept these as indicating the 
original relations, whether they imply a former state of coalescence, or of 
close approach, or of more distant approach. It is possible that the earth 
and moon were drawn together by their growth into just those relations 
which Darwin assigned to them when, in his backward tracings of their 

^ The effects of contraction are here neglected. 


history, his mathematics ceased to tell what lay beyond. At the same time, 
the hyi>othesis is hospitable to any smaller numerical values for the fric- 
tional effect of the tides which revised data may be found to imply. 

We are now prepared to inquire with equanimity what is the degree 
of trustworthiness of the astronomic data relative to the recent time- 
relations of the earth and moon. 


Near the middle of the last century Adams, from a study of certain 
data relative to the secular acceleration of the moon's mean motion, reached 
the conclusion that the earth was then losing time at the rate of 22 seconds 
per century. It is proper to add, however, that Adams laid but little stress 
on the actual numerical values which he used in computation, and that he 
was of the opinion that the amount of tidal retardation of the earth's 
rotation is quite uncertain.^ At a later date, Newcomb made a computa- 
tion based on the data then available, with the result that the rate was 
reduced to 8 seconds per century.' Darwin verified the computative part 
of Adams' results and added a neglected factor for the obliquity of the 
ecliptic and the diurnal tide which raised the estimate to 23.4 seconds per 
century. Newcomb's estimate similarly revised is 8.3 seconds.' 

A reliable answer to the question whether the earth's rate of rotation 
is or is not now departing from constancy, and at what rate, depends not 
only upon extremely refined astronomical observations, but upon the 
interpretation of these observations by means of a perfect theory of the 
lunar motions. This latter has not yet been attained. In a case where the 
suspected variation from constancy is so slight, and where the logical 
structure to be built upon it in tracing it back through tens of millions of 
years involves so great a multiplication of any error it may contain, it is 
obvious that extreme accuracy and complete soundness are necessary to 
trustworthy results. In the judgment of cautious astronomers, these 
prerequisites are not yet attainable. It is not, therefore, too much to say 
that the deductions thus far made have not a sufficiently secure obser- 
vational basis to give them authoritative value. This is not to say, by any 
means, that these results, based on the best data heretofore available, 
do not fully justify the elaborate mathematical investigations based upon 
them, for these have proved extremely illuminating and stimulative, 
and were almost necessary as precursors to the more critical work on both 
observational and theoretical lines which is necessary to give the firm 
foundation so eminently to be desired. 

> Thomson and Tait's Natural Philosophy, n, p. 419; also pp. 416-^620 and 503-^6, 
edition of 1890, and the papers of Darwin previously referred to. 

' Researches on the motion of the Moon, Washington, 1873. See also Thomson and 
Tait's Natural Philosophy, n, p. 418. 

* Thomson and Talt's Natural Philosophy, II, p. 505. 



As astronomical observations thus leave it uncertain at what precise 
rate rotation is changing at the present time, it is necessary to fall back 
upon such other evidences as the tides themselves present, and after that 
upon the geological evidences. Each of the three fundamental divisions 
of the earth, the atmosphere, the hydrosphere, and the lithosphere, is 
affected differentially by the attraction of the moon and sun, and hence 
they are all, theoretically at least, affected by the tides. They furnish a 
suggestive combination for study in that the first is a highly fluent elastic 
body, susceptible of great and easy changes of form and volume; the second 
is extremely mobile, but sensibly incompressible; while the third is solid, 
at least externally, and probably rigid as a whole and possessed of effective 
elasticity of form. Because of the markedly different properties of these 
three components of the earth, it would seem that comparisons of their 
individual responses to the differential attractions of the moon and sun 
might throw special light on tidal phenomena. 


Because the atmosphere is a highly symmetrical envelope, because its 
continuity is broken by no barriers, because it is extremely mobile, because 
it has great elasticity of volume, and because it presents greater differences 
of distance from the tide-producing bodies than the hydrosphere or the 
lithosphere, it would seem that it should give a tide of declared charac- 
teristics. We are, however, almost wholly without evidences of such a tide, 
notwithstanding the large mass of barometrical data at command. These 
data stretch over a long term of years and are refined enough to show 
several small periodic oscillations, but none of these, at least none of those 
commonly recognized, are timed with the moon. Atmospheric tides play 
no part in the science of modem meteorology. Laplace discussed the 
tides of the atmosphere briefly and theoretically and found that if the sun 
and moon were in the plane of the earth's equator and if the two bodies 
were in the same line and at their mean distances, the variation of the 
barometer would be 0.63 mm.^ Darwin, without entering upon their 
discussion, expresses the opinion that they are undoubtedly very minute.' 
Other methods of estimating the atmospheric tides support Laplace in 
showing that the amount of the forced tides should be just within the 
limits of observation, from which it is inferred that they should become 
quite appreciable if they were much reinforced by the codperation of free 
waves. The chief light which their scantiness seems capable of throwing 
on the general problem in hand is that which bears on the dependence of 
the actual tides upon the reinforcement of the forced waves by the com- 
mensiurable action of the free waves that spring from them. 

The best observational data relative to the rate of propagation of a 
free atmospheric wave arising from a forced oscillation are those furnished 

1 Mtouuque COeste, Pi. I, Bk. IV, and Bk. XIII, vol. 5, p. 337. 
> Eno. Bnt., "Tides," p. 353. 


by the great explosion of Erakatoa on August 27, 1883, as set forth by 
lieutenant-General Stracbey in the monograph of the Royal Society on the 
'' Eruption of Erakatoa and subsequent phenomena." He says: 

The observed facts dearly establish that the successive repetitions of the disturbance 
at the numerous stations, after var3ring intervals of time, were caused by the passage 
over them of an atmosphOTic wave or oscillation, propagated over the surface, of the globe 
from Krakatoa as a center, and thenoe expanding in a circular form, till it became a great 
circle at a distance of 90^ from its origin, after which it advanced, gradually contracting 
again, to a node at the antipodes of Krakatoa; whence it was reflected or reproduced, 
traveling backwards again to Krakatoa, from which it once more returned in its original 
direction; and in this manner its repetition was observed not fewer than seven times at 
many of the stations, four passages having been those of the wave traveling from Kra- 
katoa, and three those of the wave traveling from its antipodes, subsequently to which 
its traces were lost (p. 63). 

The velocities of propagation of these waves were found to vary from 
674 to 726 miles per hour — somewhat below the normal rate of sound at the 
surface of the earth, which is 757 miles per hour at 10^ C. and 780 miles at 
22^ C. The average temperature of the air at its base is 15^ C. to 17^ C, 
from which the temperature declines with ascent, as does also the density. 

The mean time occupied by the Erakatoan waves in making a first 
circuit of the earth, for the computation of which 27 stations were avail- 
able, was 36 hours and 24 minutes, the angular rate being 9.89^ per hour; 
the mean of the second circuit, for which 18 stations were available, was 
36 hours and 30 minutes, the angular rate being 9.86^ per hour; the mean 
of the last observed circuit, for which 10 stations were available, was 37 
hours and 50 minutes, the angular rate being 9.77^ per hour. 

Now if the forced tidal wave be analyzed into instantaneous impulses 
and these be regarded as discontinuous, they may each be treated as though 
they gave rise to free waves similar to those derived from the volcanic 
impulses of Erakatoa. If we compare the intermediate rate determined 
for the free Erakatoan waves with the angular rate of the forced lunar tide, 
it will appear that the latter would outrun the former at the rate of about 
4.6^ per hour. The free wave would therefore soon begin to flatten the 
surface configuration of the forced tide by extending its amplitude, and in 
less than ten hours its influence would begin to be antagonistic to the 
forced tide. This antagonistic influence would reach its maximum about 
ten hours later, but would continue with declining force for nearly another 
ten hours, beyond which, because of the relatively high viscosity of the air, 
it may be regarded as negligible. It appears therefore that the periods of 
the free atmospheric waves are not such as to effectively reinforce the 
forced waves and hence they do not rise to appreciable value. 

In addition to this there seems reason to suspect that the compressi- 
bility and the relatively high viscosity of the air may combine to cause a 
portion of the atmospheric tide to take the form of an elastic wave rather 
than of a fluidal movement; that is, the tidal force may produce alternate 
expansion and compression of the air such as would not be possible in 
water because of its incompressibility. Such expansional and compressional 
states of the atmosphere would be relieved by a prompt return to the un- 
strained condition as fast as the tidal forces were in any measure withdrawn 


and this would reduce the amount of fluidal movement on which a mass- 
tide depends. It is not unlikely, therefore, that some part of the scantiness 
of the atmospheric tide is due to the elastic constitution of the atmosphere. 
There is a semi-diurnal wave of atmospheric pressure which has its 
maximum about 10 o'clock a.m. and p.m. Lord Kelvin, interpreting 
this as an increase of mass corresponding to the increase of pressure, 
has computed that it would accelerate the rotation of the earth about 
27 seconds per century.^ If however this oscillation is merely a transient 
increase of elastic pressure at the base of the atmosphere due to basal 
heat, the ezpansional effects of which are resisted for the time by the inertia 
of the air above, as seems not impossible, the wave would have no direct 
accelerative effects on the earth's rotation. 


There is reason to suspect that the water-tides are in part derived from 
the pulsations of the lithosphere.' It will therefore be best to discuss 
these first. Since no body is absolutely rigid, and since abundant evidence 
shows that the lithosphere is appreciably yielding, there can be no theo- 
retical doubt that there are tides of the lithosphere of some kind and of 
some magnitude. The only vital questions therefore relate to their magni- 
tudes and their specific forms. 

The experimental efforts of Sir George and Horace Darwin,* of Von 
Rebuer-Paschwitz,^ and of Ehlert,* resulted in detecting only slight indica- 
tions of body tides, and even these indications were of somewhat doubtful 
interpretation. It appears, however, that the effort of these investigators 
was directed toward the detection of the general deformations of the 
spheroid directly assignable to the tidal forces, and it is not clear that the 
observed results are to be interpreted as equally adverse to the existence 
of shorter pulsations assignable to the normal vibrations of the spheroid, 
induced by the tidal strains. The nature and likelihood of such shorter 
pulsations will be considered later. 

So far as opinion as to the value of the lithospheric tides is entitled to 
weight we can not do better than to quote the conclusions of Sir George 

^ Natural Philosophy, Thomson and Tait, ed. 1890, p. 418. 

' It should be understood that this is merely an individual view unsupported by the 
expressed opinion of any special student of the tides, so to as I know, and without r&oog- 
niuon in the literature of the subject. It is based on the conviction that while the direct 
rise and fall of the surface of the lithosphere in response to attraction similarly affecting 
the water tends to reduce the amount of the water-tides, the tiltin^^ of the hthoepherio 
bed in which the oceans lie first on one side and then on the other m the course of the 
procress of the lithospheric wave must develop an inertia tide very similar to the waves 
poroduoed by the rockmg of artificial basins. It is also my view that the various free pul- 
sations that may arise fiom the forced deformations of the lithosphere may give impulses 
to the waters rating in basins on its surface and that water-waves may spring from these 
<iuite independently of the direct attraction of the tide-inxxiucing body, though of course 
indirectly dependent on it. 

* Reports to the Brit. A. A. 8. on Measurement of the Lunar Disturbance of Gravity, 
York meeting, 1880, pp. 93-126, and Southampton meeting, 1882, pp. 95-119; also 
"Tides," G. H. Darwin, 1893, pp. 108-148. 

« Das Horisontaljpendel, Nova Acta Leop. Garol. Akad., 1892, vol. 60, No. 1. p. 213; also 
Brit. Assoc. Rents., 1893; also Ueber Horizontalpendel-Beobachtungen in Wilnelmshaben, 
Potsdam, und Puerto Orotava auf Teneiifa, Astron. Nadhrichten, vol. 103, pp. 194r-216. 

*Horisontalpendel-Beobachtungen, Beitrage zur Geophysic, vol. 3, Pt. I, 1896. 


Darwin relative to their present magnitude, remarking by way of pre- 
caution that the quotations given, separated as they are from their context 
and the qualifications it carries, are liable to convey misconception of the 
author's views on points other than that for which alone they are quoted 
here, viz, the magnitude of the tides of the lithosphere. He says: 

The chief result of this paper [on Bodily tides of viscous and semi-elastic spheroids, 
and on the Ocean tides upon a yielding nucleus] may be summed up by saying that it is 
strongly confirmatory of the view that the earth has a very effective rigidity. But its 
chief "^ue is that it forms a necessary first chapter to the investigation of the precession 
of imperfectly elastic spheroids, which will be considered in a future paper. I shall then, 
as I believe, be able to show, by an entirely different argument, that the bodily tides in the 
earth are probably exceedingly small at the present time.' 

And again, at the end of the later paper referred to: 

The conclusion to be drawn from all these calculations is that at the present time 
the bodily tides in the earth, except perhaps the fortnightly tide, must be exceedingly 
■mall in amount; that it is utteriy uncertain how much of the observed 4' of acceleration 
of the moon's motion must be referred to the moon itself, and how much to the tidal fric- 
tion, and aocordin^y that it is equally uncertain at what rate the day is at present being 

It has already been made clear that Darwin's inquiry involved the 
assumption that in an earlier state, when the earth was more largely 
molten or viscous, the body tides were much greater and more effective 
than now. But if we substitute the view that the rigidity of the litho- 
sphere has been nearly what it is at present through the whole history of 
the earth, as is permitted by the planetesimal hjrpothesis, the conclusions 
quoted will apply to the whole period, with such modifications as may be 
required for differences of distance between the earth and moon. 

The substitution of an elastico-rigid earth for a viscous one affects the 
rotational influences of the tides qualitatively also. If tidal deformation 
causes a movement of the molecules of the lithosphere over one another 
in fluidal fashion, friction is the result, and the tide, under present condi- 
tions, must have a retardational influence. If, on the other hand, the mole- 
cules are merely strained elastically in their relations to one another, but 
do not shift these relations as they do in fluidal motion, the strain and 
the resilience from it act almost coincidently with the straining force, the 
original form and relations are almost perfectly restored on relaxation, 
the friction is slight, and the rotational effect will be essentially negligible. 

Now, when it is considered that a tidal protuberance, at the very most, can 
warp a line of molecules only in some such measure as 5,00^,0^0 ^ 15.00^.000 1 
it seems clear that the deformation lies far within the strain-limits of crys- 
talline rock, and probably within the strain limits of all rigid substances 
in the lithosphere. The only known substances within the outer half of 
the lithosphere that probably move as fluids under tidal stress, are the 
relatively trivial threads, tongues, or pools of lava within it, and the iso- 
lated molecules or groups of molecules here and there in the free form in 
the rigid rock. If we postulate an earth of such a degree of elastic rigidity 

^ Phil. Trans. Roy. Soc. Lond., 1879, p. 31. 

' On the precession of a viscous spheroid, etc. <Phil. Trans. Roy. Soc. Lond., Ft. II, 
1879 (1880), pp. 483-484. 


as seems to be required by the concurrent evidences of astronomical, 
geophysical, and seismic phenomena, it seems quite inconsistent to suppose 
that a brief deformation of the tidal sort can be other than a minute, highly 
distributive strain, which involves no flowage motion of the molecules 
upon one another, with the exceptions noted, and hence no friction of the 
fluidal t3rpe. There is a large body of geological evidence which seems to 
indicate that the lithosphere is able to accumulate stresses for long periods, 
which are then relieved by permanent deformations. It is difficult to 
understand how an earth could be possessed of this ability, if it yielded 
fluidally to such transient and moderate stresses as those of the tides of 
the outer part of the lithosphere. We therefore assume with confidence 
that, whatever the amount of the lithospheric tide, it is only an elastic 
strain which relieves itself almost instantly on the removal of the force 
which caused it and involves little friction. 

It does not appear probable, therefore, that the body tides of the earth, 
under this view of the earth's constitution, are an efficient agency in reduc- 
ing its rotation. 

This conclusion, however, even if fully accepted, does not appear to 
cover the entire possibilities of the case; for, even if the primary tidal 
deformation of the lithosphere has little or no rotational effect, it may 
possibly give rise to pulsations in the spheroid itself which will be com- 
municated to the water upon its surface and give rise to water-tides. If 
the periods of these pulsations are commensurate with those of the water- 
bodies arising from the direct attraction of the moon and sun, they may add 
something to these by sympathetic action, even though their independent 
value might be inconsiderable. This leads to an inquiry as to the natural 
oscillations of the spheriod and their relations to the oscillations of the 
lunar and solar tides. 


It appears to be possible to reach an approximate determination of 
the fundamental susceptibilities of the lithosphere to oscillations of differ- 
ent classes by combining the good offices of theoretical computations and 
observational inductions. The types of oscillation which need to be consid- 
ered here embrace those which traverse the interior as well as the surficial 
parts of the earth as distinct waves of propagation, and those oscillations 
of shape which affect the form of the earth as a whole. The latter are 
treated as harmonic pulsations and may spring either from the transmitted 
oscillations or from differential stresses arising from variations of attraction. 
The data relative to transmitted oscillations have been furnished chiefly by 
seismologists; the treatment of harmonic pulsations and fundamental sus- 
ceptibility to such oscillation has thus far been chiefly mathematical. 

Lamb, following earlier work by Kelvin, has shown that several different 
species of harmonic oscillations may arise from both the longitudinal and 
transverse waves transmitted through the earth.^ For a steel body of the 
size of the earth, he found the period of the slowest fundamental mode of 

1 On the vibratioDB of an elastic sphere, by Horace Lamb. kFtoc. Lond. ICath. Soe., 
vol. 13, 1882, pp. 189-212. 


oscillation which assumed the form of a harmonic spheroid of the second 
order to be 78 minutes. A series of other oscillations of lesser lengths 
would be developed. He found that the compressibility of the matter is 
not a vital factor, for if t be the time required by a wave of distortion to 
traverse the earth's diameter^ and if P be the period of oscillation of shape, 
then P ^ t 0.848 if the material is incompressible, and P — t 0.840 if 
the material preserves uniconstancy. Bromwich,^ bringing into the com- 
putation the effect of gravity, found that the gravest free period of a sphe- 
roid of the size, mass, and gravity of the earth, with a rigidity about that 
of steel, is 55 minute. The corresponding period, if the effect of gravity 
be neglected, is 66 minutes. If the rigidity be about that of glass, the 
period is 78.5 minutes if the effect of gravity be included, and 120 minutes 
if gravity be neglected. 

Nagaoka ' has made a study of the pulsations connected with the Kra- 
katoan eruptions of August 26 and 27, 1883, as recorded by the gasometer 
at Batavia, 94 miles from Erakatoa, on the supposition that these pulsa- 
tions were derived directly from the volcanic explosions and thus registered 
their relative times. He reached the conclusion that the series of eruptions 
were rhythmical with a unit-period of 67 minutes and a tendency toward 
the grouping of these shorter periods into larger ones of about 200 minutes. 
The former he interprets as an expression of the fundamental period of 
oscillation of the earth as a spheroid. Referring to the results of Brom- 
wich, he cites the coincidence of the Erakatoan periods so deduced with 
the computed periods when the assumption is made that the rigidity of 
the earth lies between that of steel and that of glass. Nagaoka also cites 
the apparent relationship of this period to seismic phenomena, and the 
apparent connection of certain of these phenomena with the Chandlerian 
nutation of the pole. 

The correspondences may be carried appreciably further. While exact 
determinations of the velocities of seismic tremors recorded at a distance 
from an earthquake are not yet available, the time required by the fore- 
most waves to traverse the earth's diameter may be taken provisionally 
at 22.6 minutes. These vibrations are generally interpreted by European 
and American seismologists as compressional waves and as passing through 
the earth along chords, or along curves of adaptation departing slightly 
from chords. The second set of tremors, generally interpreted as distor- 
tional, require about 50 per cent longer for chords up to 140^, and perhaps 
up to 180^, which would make their diametrical period about 33.75 min- 
utes. For the chords between 140® and 160% and perhaps up to 180®, 
Oldham inferred a longer period from the available observations, which 
are, however, thus far not sufficiently numerous for positive conclusions. 
These problematical vibrations may be directly transmitted or may be 
reflections. The period deduced for them is approximately double that of 
the compressional waves. The foremost large seismic waves, which have 
approximately a uniform velocity and which are interpreted as following 

^ On the influence of gravity on elastic waves, and in partictilar on the vibrations of 
an elastic globe, by T. J. A. Bromwich. <Proo. Ix>nd. Math. Soc., XXX, 1899. 
* Nature, May 25, 1907, pp. 89-91. 


the surface of the earth, pass from the point of origin around to the anti- 
podal point in about 112 to 115 minutes. The largest and strongest group 
of these large waves takes about 135 minutes. Comparing these with one 
another, it appears that the period of the maximum group of the large waves 
is six times that of the compressional waves, the former, however, travers- 
ing a semicircumference and the latter a diameter. The period of the second 
set of short vibrations for most chords is one-half more than that of the 
first set, while that of the problematic set is approximately twice that of 
the first set. Three times the period of the compressional waves, twice 
that of the best recorded distortional waves, one and a half times that of 
the problematic waves, and half that of the maximum long waves are 
each approximately 67.5 minutes, or essentially the same as Nagaoka's 
unit-period for the Krakatoan pulsations. The gasometer record of the 
Krakatoan eruptions is rather coarse and can not be read with exactness, 
but, taking Nagaoka's readings, the discrepancies between the recorded 
times of the twelve eruptions of August 27 and the periodic times on the 
67.5 minute basis are as follows in minutes: (starting-point, first erup- 
tion on August 27); -29 (—33.5, the half period, + 4.6); +14; -1.5; 
+2; +16; -3.5 (strong); + 3 (strong); +23.5 ( = 22.6, one-third period, 
+ 1); +33.5 (half period); -1.5 (the great eruption); -9 (the final erup- 
tion, strong). In the interpretation it is assumed that the eruption of 
August 26 at 6^ 20™ p.m. started a series of oscillations in the lithosphere 
which, at the end of the sixth period of 67.5 minutes, with a lag of 4 min- 
utes, had developed sympathetic relations with the volcanic forces and 
stimulated the first of the twelve eruptions that followed. These have the 
degree of correspondence to the assumed period just shown. Each erup- 
tion falling at or near the critical stage of the pulsation previously developed 
may be supposed to have strengthened the succeeding oscillations until 
the series reached a first double maximum at the seventh and eighth erup- 
tions, and a second and greatest maximum at the eleventh and twelfth. 
If this interpretation be justified, it may mean that the vibrations which 
arose from the earthquake developed into the form and periodicity of the 
fundamental vibrations of the earth-spheroid. The inadequacy of the 
data, quantitatively and qualitatively, to establish this positively is obvi- 
ous, and it may not be safe to rest much upon it; but the following are 
curiously related to it. 

The moon's synodical period, 1,490.5 minutes, is 22 times 67.75. The 
solar period, 1,440 minutes, is about 21.25 times 67.75. If 67.75 minutes 
be taken as the normal period of spheroidal oscillation, 22 of these con- 
stitute a lunar day, 314 approximately the average fortnightly excursion 
of the moon north and south of the equator, and 628 the lunar month. 
If a represent the northerly fortnightly excursion, and a' the southerly, 
each of these equaling 314 earth-pulsation periods, they will obviously 
have close commensurate relations at the periods represented by aa\ 
a (a'\'a'), and (a'\'ay, whose numerical values are 218, 436, and 870 days, 
respectively. Now, 436 days is the recent estimate of Kimura * for the 

* Fhymco-Math. Soc. Tokyo, Pt. II, 24, pp. 357-364, 1905; Sci. Obs., July 26, 1906. 
Pop. Astr., Oct. 1906, p. 469. 


larger circular element of the Chandlerian nutation of the pole, and this 
period of Eimura is perhaps to be regarded as a closer approximation 
than the earlier estimates of 427 to 430 days. As the sums of the tides 
formed when the moon is on the equator, is north of the equator, and is 
south of the equator, respectively, are different from one another, partly 
because of the differences in the moon's position and partly because of 
differences in the configurations of the lands and seas on the two sides of 
the equator, there seems to be a fair presumption that there would be a 
periodic difference in the tidal influences on the rotation of the earth 
about its axis corresponding to the fortnightly excursions, which would 
express itself in a nutation. Now if this period of forced nutation happens 
to be commensurate with the free period of the earth as a rotating body, 
the effect would be cumulative. Euler long ago computed that the period 
of free nutation of the axis of the earth, if it were an absolutely rigid body, 
would be 305 days. Newcomb, on the assumption that the earth has the 
rigidity of steel, found that the period would be increased to 447 days. 
This seems to imply that the earth is somewhat more rigid than steel 
and has a free nutation period somewhere about 427 to 436 days. As the 
fortnightly group of tides have a cumulative period commensurate with 
the latter, the nutation of 436 days may perhaps be due to the agency 
of this tidal group. 

In addition to this larger circular nutation, whose radius is about 15 
feet, there is a smaller elliptical nutation, of about 4 feet by 14 feet, with 
an annual period. This is assignable to the annual migration of the sun 
north and south of the equator, which gives rise to a variety of dynamic 
effects in the form of changes in the circulation of the atmosphere and of 
the ocean, in the accumulation and melting of snow and of ice, etc. This 
is in line with the common explanation of this minor nutation. 

It is an established principle that when the normal period of oscilla- 
tion of a body is less than the period of the periodic force acting on it, 
the oscillations of the body will agree in phase with those of the force. 
On this principle the oscillations of the lithosphere should agree in phase 
with the period of the tidal forces. There should therefore be direct co- 
operation between the waves of the lithosphere and the forced water-waves. 
On account of this close coincidence there is an obvious difficulty in dis- 
tinguishing the contributions of the lithosphere to the water-tides from 
those tides which spring directly from the attraction of the tide-producing 
bodies. The two should merge into a common tide, but, if the view here 
entertained relative to the development of water-tides through oscillations 
of the lithosphere be valid, the actual tides are to be regarded as composite. 

If the tides of the lithosphere were of the fluidal type and acted in strict 
coincidence with the water-tides, they would reduce the latter to the extent 
of their own magnitude, as urged by Kelvin and Darwin;^ but in so far as 
the pulsations of the lithosphere have the effect of a series of tiltings of the 
basins on the lithospheric surface, they must impart oscillatory movements 
to the water held in the basins. It is safe, on observational groimds, to 

' Thomson and Tait, Natural Philosophy, Pt. II, p. 439. 


aflBrm that all the oceans behave, in the main, as if they were isolated bodies 
held in basins on the surface of the lithosphere. There are no effective tidal 
belts stretching around the earth parallel to the equator and furnishing an 
opportunity for the development of a continuous tide of the canal type. 
The Southern Ocean, once regarded as such, does not prove to act in this 
way, nor do the Pacific and Indian Oceans act as a common body, as 
represented on the old tidal charts.^ For the purposes of this discussion it 
may be assumed with practical safety that the seas occupy a chain of irregu- 
lar basins linked to one another in various unsystematic manners, and that 
each of these bodies is subject, in its own way, to such oscillations as the 
rocking of its basin may impart to it. If the tides of the lithosphere are 
as small as present evidence seems to indicate, this may not be important in 
its own first effects, but as a periodic action it may become, by commen- 
surate accumulation, a not unimportant factor. Some of the peculiar fea- 
tures of the tides seem to be much more intelligible on the supposition that 
they arise from the oscillations of the lithosphere than from the direct action 
of the lunar and solar attractions. The rocking action of the basins would 
generate tides as freely on the eastern as on the western sides of the oceans, 
whereas the attraction of the moon and sun should be accumulative toward 
the western side. The tides are, however, rather higher on the eastern than 
on the western sides of the oceans. We shall have occasion to return to this 
significant feature. 

When a strain, or a deformation, or a movement of any kind is being 
impressed with increasing or declining intensity upon an elastic body which 
is already in a state of constant pulsation, as is the lithosphere, the super- 
imposed action becomes itself pulsatory, however continuous and uniform 
the increment or decrement of the superimposed action may be in itself, 
for the existing pulsation of the body alternately opposes and coincides 
with the superimposed action and gives it a corresponding pulsation. The 
water-tides assigned to the rocking of the containing basins may therefore 
be treated as composite pulsations, each advancing and each declining 
phase consisting of an undetermined number of pulsations, each of which 
gives rise to its own partial free wave. These, as do all waves of what- 
ever source, react on the lithosphere. Each such reacting pulsation, so 
far as it takes the form of a compressional wave, passes through the litho- 
sphere to the antipodes in about 22.5 minutes. It there constitutes an 
impulse acting at an angular distance of about 5.6^ in the rear of the corre- 
sponding part of the antipodal wave, tending, in its minute degree, to 
strengthen it, but with a slight increase of amplitude. The return of this 
wave requires an equal period which brings it into action at about 11.2° 
in the rear of the crest of the wave from which it sprang. This wave will 
therefore act several times in an approximately commensurate way before 
any appreciable incommensurate effects will be developed, and by that 
time its force will largely be spent. In so far as the reaction of the original 
wave develops an undulatory wave on the surface of the lithosphere, 
this wave will reach the antipodes in about 2.25 hours and will act in a 

1 See Tidal charts of the U. S. Coast and Geodetic Survey, Rept. Sept. 1900, App. 7, 
Outlines of Tidal Theory, Pt. IV a, Rollin A. Harris. Also Pt. IV b, 1904. 


similar way on the antipodal wavci tending to increase its amplitude and 
to reduce its surface gradient. On its return it will fall so far behind the 
original wave as to have appreciable incommensurate effects. The waves 
of intermediate period will have corresponding intermediate effects. In 
so far, therefore, as water-waves react upon the lithosphere and develop 
waves in it, these, while cooperating with the original waves commensur- 
ably for a time, will tend to distribute the oscillatory action into broader 
amplitudes. This may be looked upon as a tendency to develop a distribu- 
tive series of small pulsations in lieu of the original more concentrated one. 
It is observed that small pulsations attend the incoming and outgoing of 
the tides, but they have not, so far as I know, been made the subject of 
sufficient study to determine whether they are systematic or irregular, 
and whether their periods are at all in accord with the natural periods of 
the lithosphere or not. 

It is assumed in the foregoing that tidal pulsations will move through 
and over the lithosphere at the same rate as seismic pulsations, which 
probably does not involve any essential error, though pulsations vary some- 
what in their speed, even when of analogous classes. But only the general 
order of velocity is of special moment here. 

According to the mathematical investigations of Lamb, there should 
be a double series of modes of oscillation in the spheroid derivable from 
the initial impulse, of which one set should spring from the compressional 
waves and another set from the distortional waves, and these should differ 
in period. Only the gravest periods have been cited above. Without 
attempting to determine what these shorter periods are in the case of the 
lithosphere, it is probably safe to say that such as are commensurate with 
the tides of any body of water would cooperate to build these tides up and 
such as are incommensurate would have the opposite influence. Now if 
the tides of each body of water are essentially individual and are radically 
influenced by the breadth, depth, and configiuration of the water-body, 
it is not improbable that different species of both series of natural pulsa- 
tions may cooperate with the tides of different oceans and assist in their 
perpetuation and development. 

If the distribution of the strains developed by increasing or diminishing 
attraction takes place at a velocity similar to seismic vibrations, even the 
larger tidal movements of the lithosphere will act almost simultaneously 
with the tide-generating forces, for no strains of the same phase will extend 
more than 4,000 miles from the center of development of that phase. The 
extreme movement from the center to the circumference of the strained 
area would occupy, at the observed rate of compressional waves, less than 
12 minutes. In so far, therefore, as retardation of the earth's rotation is 
dependent on lag of the tide, it will be inconsequential for this class. 

Relative to the tides of an elastic earth, Darwin says: 

The other hypothesis considered is that the earth is very nearly7perfectly elastic. 
In this case the semi-diurnal and diurnal tides do not lag perceptibly, and the whole of 
the reaction is thrown on the fortnightly tide, and moreover there is no* perceptible tidal 
frictional couple about the earth's axis of rotation. From this foUows^the remarkable 
conclusion that the moon may be undergoing a true secular acceleration of motion of 


toiTwthing Imb than S.G* per century, whilst the length of the day may remain almost 

The leeuits of theoe two hypotheses (a viscous spheroid and a nearly perfectly elastic 
spheroid) show what fundamentally different interpretations may be put to the phenom- 
enon of the secular acceleration of the moon. 

Under these circumstances, I can not think that any estimate having any pretension 
to aoeuraoy can be made as to the present rate of tidal friction.^ 


If the preceding views are tenable, practically the whole tidal effect 
on rotation at present is concentrated in the water-tides. A part of these 
are assigned to the immediate action of lunar and solar attraction and a 
part to the mediate action of the lithosphere. While the lithosphere is 
thus supposed to contribute to the formation of the water-tides, this sup- 
plementary action is supposed to be qualified by its distributive action 
as previously explained. The water-tides are thus interpreted as more 
complex in origin than they have usually been thought to be. This must 
doubtless be regarded as an unwelcome infliction, for even under the simpler 
conception of their origin from direct attraction only, they are, in many 
of their phases, beyond complete mathematical treatment. These added 
complexities put them still further beyond the reach of such treatment. 
But this added complexity may, after all, only help to force us on toward 
the adoption of naturalistic methods. It has been becoming increasingly 
clear for some time that, to secure reliable results, the tides must be studied 
on a direct observational basis. The more hopeless the purely theoretical 
method becomes, the more assiduously is the observational method likely to 
be pursued. If theoretical methods are given precedence, they should be such 
as are based on the fundamental laws of energy, which hold good irrespective 
of special forms of action, however multitudinous and irresolvable. 

As already remarked, the tidal water-bodies have no systematic, much 
less have they any symmetric, distribution. Innumerable idealizations 
as to the forms and relations of the oceans have been framed, but beyond 
a few of a very general sort, they are notable principally for their imdue 
emphasis of amenable concurrences and their neglect of refractory non- 
concurrences. The north-south extensions of both the eastern and western 
continents are particularly unfavorable for the development of a con- 
tinuous forward movement of the tides. The southern ocean furnishes 
the only continuous east-west belt of ocean encircling the earth's axis of 
rotation, but, according to the cotidal charts of the U. S. Coast and Geo- 
detic Survey, this is not affected by a continuous westerly tide. The 
Pacific tides move easterly from New Zealand and, by interpretation of 
the scant data available, easterly all the way to the straits between South 
America and Antarctica, through which they move eastward and then 
northward along the Patagonian coast. The tides of the northeastern part 
of the Indian Ocean move easterly into the straits between Australia and 
Asia, while the Pacific tides enter on the opposite side and the two sets 

1 On the procession of a viscous spheroid, etc. <Phil. Trans. Roy. Soc. Lond., Ft. II, 
1879 (1880), p. 629. 


meet one another within the inter-idand water-bodies. There seems to 
be no definite perpetuation of the Pacific tides into the Indian Ocean, 
these bodies, though connected both north and south of Australia, acting 
in essential independence. While the long-prevalent view that the tides 
of the Atlantic and Arctic Oceans are derivatives from the Southern Ocean 
still has the apparent support of observational data, there are many facts 
that seem to indicate that this is only a part of the truth. A derivative 
wave should gradually die down as it progresses; but, notwithstanding 
the distance of the North Atlantic from the Southern Ocean, the tides are 
higher there than those of the South Atlantic. A derivative wave should 
be intensified in passing a constriction and should be lowered in an expanded 
water-body beyond; but, notwithstanding the reduction of the Atlantic 
breadth between Brazil and Sierra Leone, the tides are particularly high 
in the lee of the great nose of Africa north of this constriction. The north- 
easterly-trending coast of New England and the Provinces stands directly 
athwart an unobstructed stretch of sea reaching back to the Southern 
Ocean along the line of assumed propagation, and yet the average tide on 
this coast is notably less than that on the European coast of the same lati- 
tudes, though this lies behind the African projection. Comparing the tides 
on the Atlantic islands — whose isolation should render them measurably 
free from local influences, save those of their own basal slopes and their 
harbors — it is notable that the tides on the islands of the South Atlantic 
average less than half as much as those of the islands of the North Atlantic. 
The tides on the islands in the far North Atlantic, and even some of those 
in the borders of the Arctic Ocean, are singularly high, such as those of the 
Faroes, Shetlands, Orkneys, Hebrides, Iceland, Greenland, Jan Mayen, and 
some Arctic Islands of North America. On the purely derivative theory, 
these tides must be supposed to have been traveling 24 hours or more 
since they left the place of their origin, and those in the high north have 
been subject to the damping e£fects of polar ice. 

There are not a few anomalies that are very puzzling on the supposition 
of a westward drag of the waters by the moon and sim. The northeast 
coast of South America trends northwesterly and the Central American 
states continue the trend in fair alignment. Over against this, the North 
American coast has a southwesterly trend, meeting the projection of the 
northwesterly trend of South America on the coast of Guatemala, thus 
forming a wide eastward-facing angle. From this point to Cape Race, 
the angular distance is 40° and from it to Cape St. Roque, 50°, while the 
open eastward-facing mouth between Cape Race and Cape St. Roque is 
about 50°. On the hypothesis of a westward-moving tide, cumulative 
toward the west, we should expect high tides in the Antilles, the Caribbean 
Sea, and the Gulf of Mexico. If it is thought that the last two bodies are 
protected from this high tide by the Antilles, the tides on the eastern side 
of the Antilles should be markedly high. The record does not show this. 
The tides on the African and European coasts opposite are notably higher 
than those which might be supposed to be unusually concentrated within 
this angle. As bearing on any supposed protection of the Gulf of Mexico 
and the Caribbean Sea by the Antilles, a comparison may be made with 


Hudson Bay, which is far more land-locked and is in a much higher latitude 
and much farther from the assigned source of derivation. Singularly 
enough, the tides in Hudson Bay are several times as high as those of the 
Mezioan Gulf and the Caribbean Sea. 

If it be objected that the North Atlantic is too narrow and too peculiar 
in its relations to give these singular features much weight, a similar line 
of inspection may be applied to the Pacific, whose breadth and equatorial 
position make it preeminently favorable for a westward accumulation of 
the tide. From the mass of data now made available by Harris's compila- 
tions and harmonic reductions, it appears that the tides on the eastern 
side of the Pacific as recorded on the American coasts are notably stronger 
than the average tides on the Asiatic coasts. A comparison of the tidal 
heights on the Pacific islands, though the data are inadequate, also fails to 
show a concentration on the western side. 

On the old cotidal charts, and more definitely on the new ones of Harris 
(fig. 3), it appears that the dominant tide of the Pacific originates in a 
singular loop near the Galapagos Islands, off the coast of South America, 
from which, on one side, a wave moves easterly and southeasterly to the 
South American coast in strong force, then down it to the extremity of 
the continent, where, according to Harris, it rounds Cape Horn and moves 
up the eastern coast of Patagonia to about the mouth of the Rio de la 
Plata. On the other side of the Galapagos loop, a wave moves north- 
westerly along the North American coast, and then westerly toward the 
Asiatic coast. In the heart of the Pacific, Harris locates three amphi- 
dromic centers of practically no tide (fig. 3). While these are not directly 
based on observations, they are believed by this industrious and original 
student of the tides to be in accord with the data derivable from the observed 
tides of the central Pacific Islands. This singular dispersion of the Pacific 
tides from the vicinity of the Galapagos Islands near the eastern border 
of the great ocean, as shown on the old charts, suggested to me, perhaps 
a decade ago, that the water-tides might be largely derived from the litho- 
sphere rather than directly from the attraction of the moon and sun on the 
water itself. The rocking of the basins, first by a lift on the east side and 
later by a lift on the west, under the progressive influence of the tide- 
producing body, seemed to me more compatible with this behavior of the 
tides than direct attraction on the water itself, which I supposed should be 
less effective on the east side than on the west. The greater strength of 
the tides on the east side of the Atlantic also strengthens the impression 
that, whether this suggestion of derivation from the pulsations of the 
lithosphere be of any value or not, the actual evolution of the tides involves 
much more than the simple upward pull and westward drag of the waters 
by the moon and sun. 

The recent theory advanced by Harris,^ that the tides are largely due 
to the cumulative agency of stationary oscillations in such segments of 
the oceans as may act commensurately with the tidal forces, goes far to 
relieve the foregoing and similar features of the tides of their seeming 

t "Outlines of tidal theory/' Rollin A. Harris, Rep. U. S. Coast and Geodetic Sunr. 
1900, app. 7, pp. 636-e09. 


incompatibility with the theory of direct attractional action on the water 
itself; indeed, the concrete features of Harris's theory seem to have been 
developed largely by a study of these remarkable features, and to be an 
attempt to give them a dynamical expression in terms of the direct-action 
hypothesis. Whatever shall be the final judgment regarding particular 
aspects of this theory, whose author claims for it only a partial explana- 
tion of the tides, it seems eminently probable that commensurate oscilla- 
tion is a vital factor in building up the waves of the actual tides. The 
attempt to work out a concrete theory of the tides from their specific phe- 
nomena is greatly to be commended, for by concrete application alone are 
the proximate sources of the tides likely to be determined. This may be 
said without derogation of the value of the more general theories. 

Concerning the insufficiency of simple attraction without sympathetic 
intensification to explain the actual tides, Harris says: 

In approaching the question of the actual causes of the tides, upon which so much 
labor has been expended and oonceming which so much has been written, one may well 
surmise that the subject does not admit of accurate or complete treatment. It is there- 
fore natural to consider, in the first place, only those sources which would seem to account 
for the dominant tides in any given region under consideration, and to postpone, perhaps 
indefinitely, the consideration of those sources whose importance in the production of the 
tides must be relatively small. Considering the actual distribution of land and water a 
few computations upon hypothetical cases will suffice to convince one that as a rule the 
ocean tides, as we know them, are so great that they can be produced only by successive 
actions of ibe tidal forces upon oscillating systems, each having as free a period, approxi- 
mately the period of the forces, and each perfect enough to preserve the general character 
of its motion during several such periods were the forces to cease their action.* 

In another place he says: 

Unless the free period of a body of water, or of some portion of this body, approxi- 
mately agrees with the period of the tidal forces, the tide in the body proper must be small, 
and generally smaller than the theoretical equilibrium tide for the body in question. But 
in many parts of the oceans, the tide is several times greater than that which could be 
raised by the forces, even if we could suppose sufficient depths and sufficiently complete 
boundaries for enabling equilibrium tides to occur. Hence regions the dimensions of 
which approach critical values must exist in the oceans and account for the principal tides. 

That stationary oscillations of unexpectedly large amplitude exist in the oceans there 
is abundant evidence. In fact, a glance at the charts will show regions of large ranges 
over each of which the time of the tide varies but little. As a nodal line is approached the 
range diminishes, and the time of the tide changes rapidly in a comparatively short distance. 
Moreover the dimensions of the oceans are such that areas having nearly critical lengths 
can be readily discovered; these respond well to the forces, and their tides must be the 
ruling semi-diurnal tides of the ocean.' 

Harris has attempted to detect those portions of the oceans whose 
lengths, depths, and relations make them susceptible to the development 
of free oscillations whose periods are sufficiently near to the periods of the 
tidal forces, or to some simple fraction of them, to permit cooperation in 
building up effective stationary systems of oscillations. Of the major 
order, he finds a northern and a southern system in the Indian Ocean, a 
South Atlantic system, a North Atlantic system, and two systems in the 
Pacific, as well as a large number of systems of the minor order. Diagrams 
and details of the main systems are given in the original paper. 

^ Loc. cit., p. 624. ' Nature, Feb. 22, 1906, p. 388. 


Harris does not claim that the demonstration of these is complete 
or finali and he recognizes that much additional data will be required for 
a full verification of the postulated systems and for completing the full 
cat^ory of systems, but he believes that those announced correspond 
fairly well with existing knowledge. Question has been raised as to whether 
the phases of oscillation which he assigns to his systems are such as would 
naturally arise from the forced waves. Question has also been raised as 
to the adaptability of the oceanic segments to oscillate as postulated.^ 
Harris contributes a new map of cotidal lines for the world (fig. 3) in which 
interpretation is conveniently combined with the observed data which 
are also given separately in tables and in sectional charts. By comparison 
of this cotidal map with previous charts, it will be seen that Harris's inter- 
pretation departs rather markedly from that implied by the previous 
cotidal charts. 

A significant feature of Harris's cotidal map is its amphidromic nodes, 
centers of little or no tide about which the tidal wave swings in the course 
of the twelve-lunar-hour period. Three of these amphidromic points are 
located in the Pacific, one in the North Atlantic, and one in the Indian 
Ocean (fig. 3). These points are associated with nodal lines that separate 
the oscillating sections of the systems to which they are assigned. That 
there should be such nodes of little or no tide in the heart of the great 
oceans, where under the familiar mode of interpretation the tidal waves 
should have their freest sweep and greatest strength, well expresses the 
extent of Harris's interpretation^ departure. 

It appears then that, under the broad mantle of the postulate that the 
tides are due to the attractions of the sun and moon, there are three special 
or proximate views as to the immediate origin of the actual tides: (1) the 
direct attraction of the tide-producing bodies on the water; (2) the effect 
of stationary oscillations promoted by such direct attraction; and (3) the 
to-and-fro tilting of the rock basins in which the water-bodies rest by 
the tides of the lithosphere. The older view has always recognized the 
supplementary effect of the natural oscillations of the water-bodies, but it 
has never given them a prominent place nor quite that distinctive form 
which has been assigned them by Harris. Since the cooperation of oscilla- 
tions is independent of their source, any waves that may come from the 
lithosphere are as available for building up systems of stationary oscilla- 
tions as are those springing from direct attraction, since they are likely to 
be timed quite as well. Harris's theory, or any theory of its kind, may 
therefore find as good a working basis in tides derived from the lithosphere, 
so far as these go, as in those formed by direct attraction. 

The vital question here is the bearing of the deductions from these three 
points of view — assuming that each of them represents some truth — on 
the rotational problem. Let us first consider this on the familiar assumption 
that retardation is dependent on the position of the waves, assuming that 
to produce retardation the wave on the moonward side must be in front 
of that body pulling it forward and being itself in turn pulled backward 

^ Nature, Sept. 4, 1902, p. 444; Apr. 23, 1903, p. 583; Jan. 11, 1906, p. 248; Feb. 22 
1906, p. 388. 



over the earth's surface (figures 1 and 2). Afterwards let us consider 
the phenomena from a more radical point of view founded on the laws of 
energy and the configuration of the interacting bodies. 

From what has already been said, it is clear that no continuous tides 
are being dragged around the earth acting as a frictional band. No single 
tide moves westerly so much as one-half of the earth's circumference, and 
most of the tides have a much less movement in that direction. On the 
other hand, many tides move easterly and still others move northerly and 
southerly. The position of these relative to the moon is various, and the 
attraction of the moon upon them may be accelerative or neutral as well as 
retardative, so far as instantaneous attraction while in the given positions 
is concerned. A wave starting in the Southern Ocean and moving north 
through the Atlantic for more than a day will rim the whole gamut of 
positional relations to the moon and sun, and will, considered simply as an 
attached protuberance, be retardative, neutral, and accelerative in turn. 
In the case of waves that move to and fro across the water-bodies in seiche- 
like fashion, it is obvious that the positional relations may be various. 
The most interesting cases are those of water-bodies whose periods of 
oscillation are nearly commensurate with the periods of the tidal forces. 
The breadth and depth of a water-body may be such that a wave started 
under the moon when it passes over the eastern margin will cross to the 
western side and return to the eastern just in time to fall imder the moon's 
next crossing of the eastern margin, and so be reinforced by every return. 
In a body a little wider or a little shallower, the return of the wave would 
fall behind the moon's arrival and at its turn tend to retard the moon's 
motion, while in a body a little less wide or a little deeper the turn will 
come before the moon's arrival and the wave, at its turn, will tend to 
accelerate the moon's motion. But if either of these waves were to be 
followed through its whole course and its relations to the moon observed, 
it would be found to be accelerative, retardative, and neutral at different 

Pursuing this linQ of inspection, it may be seen that the waves developed 
in the basins of the lithosphere must have a wide range of periods, some 
longer, many shorter, than the period of the tidal forces. Their rotatory 
influence on this basis of treatment is thus extremely difficult to analyse 
and evaluate, and the algebraic sum of all such influences is quite beyond 
mathematical determination. 

The case is even more complicated when we consider amphidromic 
systems and those whose oscillations lie in lines oblique to the axis of 
rotation and to the moon's course. In the case of a stationary oscillation 
neither forward nor backward drag seems to be predicable as a total result, 
on this basis of treatment. 

When all of the multitudinous phases are considered, it is clear that 
the case becomes so extremely complex that' it can not be solved with any 
assurance of a reliable conclusion by analyzing the rotational effects of 
individual cases and summing the results. Some more basal method, so 
chosen as to escape these complications and the uncertainties of their 
interpretation, is required. 


Poincar^y after a mathematical treatment of the influence of the water- 
tides on the earth's rotation in the endeavor to simplify the case, reached 
the following conclusion: 

L'infliienoe des mardes oo^aniennes sur la durde du jour est done tout k fait minime 
et n'sit nuUement comparable k I'effet dee mar^ dues k la viscosity et k T^lasticitd de la 
pariie aolide du g^be, efifet sur lequel M. Darwin a insiBt^ dans une s6rie de M^moires du 
plus haut int6r^.* 


To the foregoing method of treating the rotational effects of the tides 
on the basis of the poaitiona of the tidal protuberances and depressions, 
as such, there seem to be, as previously intimated, graver infelicities than 
those of mere complexity. The method appears to be defective in neglect- 
ng the cooperating effects of the changes of kinetic and potential energy 
that are associated with these differences in the distribution of matter. 
These protuberances are not fixed masses of matter, but rather aggregates 
of variations in the paths of the molecules of water in their revolutions 
about the earth's axis. In the production of these protuberances and 
depressions there are reciprocal increases and diminutions of the potential 
and kinetic energies of the water particles involved. In analyzing the 
influences of these on rotation, it will be serviceable to separate the factors 
of inertia and friction — ^including under the generic term friction all obstruc- 
tive effects growing out of the relations of one particle to another — because 
the fimctions of these factors are contrasted, since the inertia tends to 
perpetuate any given state of motion, while the friction tends to reduce the 
amount of motion. There are also certain advantages in considering each 
particle separately as a body in revolution about the axis of rotation. 

Let therefore the lithosphere be regarded as a perfect spheroid sur- 
roimded completely by an ocean of uniform depth, and let the matter of 
each particle be regarded as concentrated into a point and separated from 
its fellow particles by a complete vacuum, but let the collapse of the particles 
be prevented by a hypothetical force taking the place of the resistance to 
condensation which affects the water in nature. We shall then have an 
ocean made up of mass-points which move in perfect freedom from frio- 
tional and other obstructive relationships; in other words, these points 
will constitute satellites of the lithosphere which may here be regarded as 
a rigid body acting as a massive point at its center. The behavior of the 
mass-points may then be treated, qualifiedly, according to the principles 
of celestial dynamics. In fig. 4, let E represent the earth, the circle L the 
surface of the lithosphere, and the circle A BCD the ideal surface of the 
hydrospheric satellites when revolving without perturbation by the moon. 
Then, according to the principles of celestial mechanics, first applied to 
this dass of cases by Newton,' the orbit of a particle, p, will be a closed 
curve, abed, closely resembling an ellipse, whose major axis is transverse 
to a line joining the centers of E and M. The general configuration, it 
will be noticed, is that of the ''inverted tides." The particle p in passing 

> Bulletin Aatronomigue, vol. 20 (1903) " Sur un Thter^me Q^n^ral Relatif auz Mar6M/' 
par M. H. Poincar^, p. 223. 

' Mouhon'B OdestiA] Mechanics, art. 156, p. 243. 



from a to b will lose velocity — and hence kinetic energy — and gain potential 
energy. At the point b it will have the minimum of motion and the maxi- 
mum of potentid energy. From b to c it will fall back toward the center, 
a portion of its potential energy being converted into kinetic, and its veloc- 
ity being increased and reaching a second maximum at c. In the second 

half of its orbit, cda, similar 
exchanges of kinetic and poten- 
tial energy will take place. 

If p is affected by no friction 
or obstruction in its course, 
these exchanges of kinetic and 
potential energy will be com- 
pensatory and maybe continued 
indefinitely without affecting 
the rotation of E. The case is 
that of an inner satellite or an 
inner planet when all the bodies 
involved are considered as rigid 
bodies or massive points. But 
if now friction be introduced 
at any paint in the orbit of p, 
heat wiXL be developed and 
dissipated, and energy lost to 
the system. Looked at in 
detail, it would seem that the 
retardation of p by friction on 
E in some phases of its orbit 
would be accelerative to E'a 
rotation, and in other portions, 
retardative, for in some por- 
tions p'a angular motion is 
greater than E'a, and in others 
less, but traced out in its full 
history it appears that what is 
seemingly accelerative in one 
phase is retardative in another 
and that the ulterior effect is 
precisely measured by the loss 
of mechanical energy by con- 
version into radiant energy and 
dissipation. For example, fric- 
tion between p and the normal 
periphery of J? as represented 
by A BCD, at or in the vicinity of e, will be accelerative in its immediate 
phase because p in this part of its orbit is moving faster than the contact 
portion of E, but the retardation of p in this portion will reduce the rise 
of p in the section at and near d which is retardative in its immediate 
phase because in this position p is moving more slowly than the normal 


periphery ABCD. These would be precisely compensatory — outside action 
n^eeted — ^if there were no loss of energy by dissipation. This loss reduces 
the mechanical action of the system by precisely the mechanical equivalent 
of the heat lost. 

In the actual system the distribution of the loss of energy is necessary 
to a complete solution. When the earth-moon tides alone are considered, 
it is clear that the lost energy must come either from the rotation of the 
earth or the revolution of the moon, or from both partitively. Now the 
moment of momentum of the system must remain constant and the distri- 
bution of the loss of energy must be such as to meet this requirement. 
Just what this distribution must be in a given case depends on the con- 
figuration of the system which then obtains, and is a question of celestial 
mechanics rather than of tidal theory. For the purposes of the present 
discussion it is of decisive moment to know whether the configuration of 
the earth-moon system at present is such that the earth and moon may 
recede with reduced rotation of the earth if the dominant phase of the tides 
is what has been called retardative, or may approach with accelerated 
rotation of the earth if the phase is what has been regarded as accelerative, 
or whether the configuration of the earth-moon system is such that it can 
move in one direction only when loss of energy by tidal friction takes place, 
bearing in mind that all tides give rise to friction and dissipation of energy. 
For a solution of this I appealed to Dr. Moulton, who found that, under 
the present astronomic relations of the earth and moon, any loss of energy 
by their interaction requires that the bodies recede from each other and 
that the rotation of the earth be diminished. This he finds to be rigorous 
under the laws of energy, and it seems to follow as a necessary inference 
that the special phase of the tidal action which caused the loss of energy 
is immaterial. While this determination at the time was wholly indepen- 
dent, it was soon recalled that Sir George Darwin had made a similar deter- 
mination so far as the dynamic relations of the earth and moon under loss 
of energy are concerned. It does not appear, however, that he drew the 
inference we have just drawn, for this seems to exclude any differential 
e£Fect dependent upon the positions or phases of the tides, other than such 
as is expressed in the amount of friction involved, which is dependent 
solely on the amount of the water-movement and not on its phase. 

There are assignable configurations of the system in which a movement 
in the opposite direction would take place if energy were lost by interaction. 
Such a configiuration would obtain if the centers of the earth and moon 
were within 9,000 miles of each other or their surfaces about 4,000 miles 
apart. The precise figure reached by Dr. Moulton, all computable influ- 
ences being taken into account, is 9,241 miles. Dr. Lunn computed this 
independently, using slightly different values for some of the factors — essen- 
tially those of Darwin — and neglecting some inconsequential ones, and 
foimd the distance 9,113 miles. From this the conclusion seems inevitable 
that if, for any reason, the moon were separated from the earth within 
this distance, the tidal interaction of the earth and moon would tend to 
bring them together, an adverse tendency which the fission theory must 
face in addition to those cited before. 



Relative to the question whether the lost energjr, in the present con- 
figuration of the earth-moon system, must oome wholly from the rotation 
of the earth or in part from the revolution of the moon. Dr. Moulton finds 


that the loss would be derived partitively in the ratio of 27.3 from the 
rotation of the earth to 1 from the revolution of the moon.^ 

*The general relations here involved are shown graphically by Sir Georse Darwin 
in Thomson and Tait's Nat. Phil.. II, p. 511. These are developea in much deti^ and with 
many api^cations in the paper ot Dr. Bfoulton in this series, some of the details and appli- 


W Now, it being thus rigorously shown that the energy dissipated by the 
reaction between the tide-producing bodies must be taken from the mechan- 
ical energies possessed by the bodies, respectively, in the proportions 
given, and it being further shown that the evolution must take this direc- 
tion and no other, it is clear that it will be possible to determine the effect 
of the water-tides on the earth's rotation if we can estimate the total 
energy dissipated by the tides, for the proper proportion of this can then 
be subtracted from the kinetic energy of the earth's rotation. We can of 
course use only imperfect data at present, but the uncertainties of these 
can be covered by allowances so as to approximate the true order of mag- 
nitude, and if the value does not prove to be a critical one, the order of 
magnitude may be as decisive for all geological purposes as a precise deter- 
mination. If the value proves critical, the serial method may be applied 
to make the results cover the whole range of uncertainty. By thus dealing 
with the whole of the friction, and by applying it by means of the rigorous 
laws of energy, we not only avail ourselves of a radical mode of treatment 
but avoid the tangle of special interpretations and the discrepancies of tidal 

Preparatory to an attempt to compute the total friction, it is worth 
while to note that an altogether exaggerated impression of the friction 
of the tides is inevitably conveyed by their association with wind-waves, 
river-currents, sea-currents, and other water-movements. As all of these, 
or nearly all of these, are the products of energy communicated to the earth 
by the sun's radiation, they may be assumed to be neutral in their rota- 
tional effects. If all these adventitious elements be removed in imagination, 
and the hydrosphere be made to take on a perfect calm, save as affected 
by the tidal forces, the picture of the frictional effects will be radically 
transformed, as may be seen by simple inspection. In the mid-ocean a 
water-particle will merely describe a circuit of a few feet in twelve hours, 
its movement on its fellow particles which are pursuing a somewhat similar 
circuit being almost imperceptibly slow; the movement on the bottom will 
be very slight. On a shore shelving at the rate of 1 foot in 50 feet, and 
with a tide of 6 feet, the edge of the tide would advance 250 feet in about 
6 hours, or a little over 40 feet per hour. On the exceedingly low slope 
of 1 foot in 800 feet, the advance of a 5-foot tide would be only 660 feet 
per hour. There are of course concentrations of motion in bajrs, straits, 

cations of which were woiked out after this was written. In the acoompanyinc graphic 
fllustration by Dr. A. C. Lunn, the relations of the factors are somewhat (ufferently 
arranged. In the upper diagram, fig. 5, the line " O = Af | " represents the angular velocity 
of the rotation of the earth; the une, «, the angular velocity of the moon in its orbit; r, the 
Hi'g^ni^ between the centers of the earth ana nu)on; N, the relative days in the month; 
X = Mm, the orbital moment of momentum of the system. In the lower diagram, E repre- 
sents the total energy of the earth-moon system; Ki, the kinetic energy of the earth's 

The crossing of O and «. where the angular velocity of the earth's rotation and the 
itngiiUr velocity of the moon's revolution are eqiial and they move as though a rigid body, 
is somewhat over 9,000 miles from the line of reference at the left. From this point of 
orosnnff to the left, where the distance of the centers is declining as shown by r, the total 
energyls declining as shown bv J^ in the lower diaipam and loss of energy promotes move- 
ment to the left. To the right of the crossing of O and u the centers move apart, the 
total energy declines, and loss of energy promotes movement to the right. 


and other special situations, that give quite notable movements, bat the 
proportion of these to the whole is eadly exaggerated. From a simple 
inspection of this kind it may be seen that the actual movement of the 
water, as distinguished from the wave-form, is of a low order and hence the 
friction is relatively small. 

If one were to entertain the thought that the energy of movement of 
the tides arrested by the continents is a factor of much value in rotation, 
it should be recalled that the arrest of the easterly moving tides is to be 
set over against that of the westerly moving tides, and in the case of tides 
moving obliquely to the earth's axis there are easterly components to 
offset the westerly components. We have previously noted that the tides 
on the eastern sides of the Atlantic and Pacific average higher than those 
on the western side, so that, on the face of things, the balance of influence 
would, if the reasoning were fundamentally sound, favor rotational accel- 
eration. But an analysis of the action shows that it results chiefly in a 
return wave, and this throws the problem back upon the dissipation of energy 
through friction. Some part of the energy of impact is not recovered in 
the return wave either through elastic resilience or the increased head which 
actuates the tidal ebb, but if the return wave is computed as though it were 
equal to the advance wave the loss will be covered. 

Hough has made an important contribution to the frictional phase of 
the tidal problem by an investigation of the influence of viscosity on tidal 
waves and currents,^ in which he included stationary oscillations as well 
as progressive waves. He considered separately the tides whose wave- 
lengths are large in comparison with the depth of the water-body and 
those in which the wave-lengths are small. If attention were confined to 
the main tidal phenomena, it might be thought that the wave-lengths 
small in comparison with the depth of water would have no representatives 
in tidal action, since the greatest depths of the sea are small compared with 
the amplitude of typical tidal waves, but, on consideration, it is apparent 
that these waves of great amplitude, by their various interactions upon 
one another, by the modifications which they suffer in their approaches 
to the shores, their arrests and their retreats, as also by their multitudinous 
interactions with wind-waves, sea-currents, etc., give rise to an indefinite 
number of waves of lesser amplitude, and that the length of these second- 
aries may be short compared with the oceanic depths. Hough did not 
consider the special effects of the irregularities of coasts and some other 
modifying conditions and these must be recalled in interpreting his extra- 
ordinary results. When these various agencies of modification are duly 
considered, it seems probable that after a very few days the original tidal 
oscillations largely lose their primitive amplitudes and take on shorter 
ones, and that in this way they pass beyond observation long before they 
actually die out. 

Hough did not attempt to determine the complete destruction of the 
tidal waves by friction, but merely the time required for their reduction 
to the value represented by lie, or 0.368 of their original value. He found 

> Proc. Lond. Math. Soc., vol. 28, pp. 264-288. 


that for waves of the class corresponding in his terminology to n^^O, in 
waters of abysmal depths, the modulus of decay (to 0.368 value) is 42.6 
years; for the class corresponding to n~l, 31.7 years; while for that 
corresponding to n = 100, it is 4.8 years. The t3rpes, n«=0, n = l, represent 
the longest and strongest waves; while those in which n has a high value 
represent waves of much shorter and feebler type and hence those first 
to be reduced to the limit named, as indicated by the moduli. In a case 
where the depth is 200 meters — the prevailing depth on the edge of the 
continental shelf — the moduli of decay for the types n«>0 and n« 1 are 4.5 
and 3.3 years, respectively. For waves of 100 meters amplitude in water 
with a depth of 1 meter, the modulus of decay is about 80 minutes, bottom 
friction included. If there were no friction at the bottom, the modulus of 
decay would be about 2.25 years. In summation Hough says: 

These results indicate how little can be the effects of viscosity upon the motion of 
the sea, except possibly in usually confined waters. It seems that wherever the depth 
exceeds a very moderate amount, say 100 fathoms, the rise and fall of the waters due to 
the sun and moon will not be appreciably affected by friction.^ 

These determinations have a significant bearing upon the question how 
much of a given tidal wave is due to the force that has just been acting 
upon it dming the current tidal period, and how much to the residual 
motion inherited from previous tides. If the motion of the waters when 
once generated requires these long periods for subsidence, it is obvious 
that each tidal wave may be perpetuated so as to codperate with a long 
series of forcing actions in succeeding periods, if its period is commensurate 
with these. This supports the view, previously discussed, that the waves 
observed in those portions of the ocean most favorable for sympathetic 
accumulation are the products of a considerable series of forcing actions. 
This means that, at least in such cases, the element added with each tidal 
period is not measured by the actual waves observed, but by some minor 
fraction of it, and hence that the tidal friction which is daily exerted on 
the earth is by no means the amount necessary to reduce the observed 
tidal movement to zero, but merely that which is necessary to offset the 
daily increment, or its equivalent, the daily factor of decay. A wave of 
the type n»0 in water 200 meters deep, if commensurate action were per- 
fect and dissipation, by giving rise to derivative waves, were wholly absent, 
would need to have less than ^-^ of its value added to it daily to maintain 
its value. This must not be taken as representing an actuid case, but it 
appears from, considerations of this kind that the total energy of motion 
expressed by the tides daily is by no means a safe basis for estimating 
the energy lost through them. This must be computed directly from the 
water^movements under the conditions that actually affect them. 

There is a check on carrying considerations of this kind too far, in the 
fact that the spring and neap tides and other special tides that depend on 
variations in the relations of the tide-producing bodies pass through their 
climacteric phases within one or two days of the astronomical configura- 
tions that give origin to them. They increase and die away with a relative 

» Loc. cit., p. 287. 


promptness which shows that no tidal agency perpetuates its speeial eflfeots 
in a special phase for a very long period. 

There is a minor qualification of Hough's results that is worthy of 
passing notice. He took for his coefficient of viscosity 0.0178, which is 
one of the determinations for 0^ C. The viscosity of water is much in- 
fluenced by temperature. The coefficient for 17^ C. is 0.0109. As this is 
about the average temperature of the surface of the earth, it may be taken 
as roughly, though not accurately, the average temperature of the ocean 
water. Salinity increases the viscosity. While I do not know of any direct 
determinations on sea-water, the determinations for normal solutions of 
sodium chloride imply that the viscosity of sea-water at 17^ C. would be 
about 0.012, which makes Hough's results very conservative, so far as aflfeoted 
by the coefficient of viscosity used. This, however, is not so much the point 
as is the difference in the viscosity of the bottom and the surface water, 
respectively, in the low latitudes, that of the deep water being somewhere 
about 0.0195, while that of the surface in the tropics is about 0.0099, 
or but little more than half as great. This difference must increase rela- 
tively the motion of the water on itself and reduce that upon the bottom. 
It will thus substitute a distributive movement within the water, in which 
the friction is very low, for a more concentrated movement between the 
film of water attached to the bottom and that immediately above it, in 
which the friction is relatively high. The total effect is to reduce the frio- 
tional value. 

Radical as are the suggestions of Hough's inquiry, a way to apply them 
directly, so as to secure a numerical expression of the total vaJue of the 
friction of the tides in terms of work done, has not been found, and I have 
therefore tried to shape the problem so that it could be treated by the 
method of the engineer, as a given mass of water, with a given amount of 
flow, in a given time, under assigned conditions. To do this it seems neces- 
sary to substitute for the actual ocean an equal body of water in a more 
tractable form, but subject to equivalent friction. In doing this I have 
endeavored to give to the movements of the substitute ocean at least as 
great friction as that of the actual ocean. For the purposes of frictional 
treatment the ocean may be regarded as consisting of three portions, (a) 
the shallow water between the coast and the edge of the continental shelf, 
(b) the water on the slope between the edge of this shelf and deep water, 
and (c) the deep portion. 

(a) The coast line is taken at 120,000 miles, which is nearly double the 
simple outlines of the ocean and about 5 times the earth's circumference. 
The depth of water on the outer edge of the continental shelf is taken at 
600 feet, the accepted depth. The water is made to deepen uniformly 
from the coast to the edge of the shelf, which very greatly exaggerates its 
shallowness near the shore, where the friction is relatively greatest. The 
area is taken at 12,000,000 square miles, or 20 per cent more than Murray's 
estimate for this part of the actual ocean. 

(b) The portion b is given a width of 50 miles, a descent from 600 to 
9,000 feet, and an aggregate length of 120,000 miles. This length is much 
greater than the actual length of the continental margin, the excess being 


intended to include a liberal allowance for the slope-tracts of the oceanic 
islands, which, however., are not subtracted from area c. 

(c) The abysmal section is given a depth of only 9,000 feet and an 
area 20 per cent greater than that of the deep ocean, this reduction of 
depth and increase of area being intended to offset the frictional effects 
of the inequalities of the actual bottom. 

The mean height of the tide in the substitute ocean is taken as 4.09 
feet, which is equivalent to 4.9 feet for the actual ocean. The mean range 
of the tide for the 280 available stations given in Harris's table of tides 
harmonically analysed is 4.548 feet.* As these stations are chiefly in harbors 
where local concentration is felt, 4.548 feet is probably rather high for the 
average range of the tide, even on the coasts, and it is certainly much too 
high for the mean range over the whole ocean. In using the equivalent of 
4.9 feet for the substitute tide in addition to the large allowances made 
above, it would appear that the computation is amply guarded against 

In using the foregoing guards against underestimation, which seem to 
me excessive, I have been somewhat influenced by the thought that there 
are derivatives from the observed tides which are not recognised and 
measured as such, but whose dissipation of energy should be covered by 
the computation. But, however well guarded, it is not presumed that 
any results now attainable will have much value beyond indicating the 
order of magnitude of the total friction. With the foregoing precautions 
the results should not be seriously less than the actual fact. But, if they 
are thought to be so by any one, the results can easily be multiplied accord- 

With these data, a computation was made by Dr. W. D. MacMillan 
in the manner set forth by him in a following paper of this series, p. 71. 
This computation, it will be observed, was made for continuous motion, 
but in estimating the rate, 12.5 minutes between each lunar tide were 
allowed for the turn of the tide. He finds the yearly loss of energy to be 
38,918 X 10*' foot-pounds. The rotational energy of the earth, reckoned 
on the assumption that the Laplacian law of density obtains, is 157 X 10^ 
foot-pounds. At the computed rate of loss, this amount of energy would 
last 40,440,000,000 years. The length of the day would be increased one 
second in about 460,000 years. In 100,000,000 years the total lengthening 
of the day would be about 3.6 minutes. 

If this result does not wholly misrepresent the order of value of the 
friction of the water-tides, it follows that, even if the allowances for the 
irregularities of the tidal water-bodies be greatly increased, and if the 
formulas of the engineers for the effects of friction be multiplied several 
times, and other allowances be made in the most generous manner, the 
effects of the water-tides on the rate of rotation of the earth during the 
known geological period are negligible. If the friction of the body-tides 
and the air-tides is also very small, there is no reason to expect to find in 
the geological evidences any appreciable deformations of the earth's body 
the distinctive characteristics of tidal effects. On the other hand, 

' Rept. Coast and Geodetic Surv., 1900, pp. 664-677. 


if the deductions which have heretofore been drawn from the older eos- 
mogonic and geophysical conceptions are true, there should be geological 
testimony to support them. We turn therefore to the geological evidences 
with heightened interest. 


Perhaps the most important of the geological lines of approach to the 
rotational problem, is found in the evidences of an appropriate change 
or lack of change of the earth's form. At least it is this problematic change 
of form that gives the subject its obvious importance in diastrophism, to 
which this discussion is a preface. If the rotation of the earth were once 
appreciably faster than now, either the form of the lithosphere would have 
been more oblate than it is at present, or the surface-waters would have 
been accumulated at the equator by the increased centrifugal force, or 
both actions would have taken place conjointly; and a change from this 
configuration to the present one must have followed. If the lithosphere 
has changed its form appreciably within known geological times owing to 
reduction of rotation, such a change should be manifest in its structural 
deformations, especially in the deformations of the early ages. If the 
lithosphere has not essentially changed its form because of reduced rota- 
tion, but the waters served as the accommodating factor, this, if it were 
of sensible amount, should have been manifested by deposits of the kinds 
that imply prevalent and deep submergence in the equatorial regions 
and by erosions signifying prevalent and pronounced emergence in high 
latitudes in the former ages of higher rotation, and by the reverse in the 
later ages, both of which would be shown by the geological records of those 


The bearing of a possible change of form, assignable to a change of 
rotation, on terrestrial diastrophism has long been recognised in some 
measure by geologists, but the first attempt to reduce it to numerical terms 
seems to have been that made by President Van Hise several years ago.^ 
He inspired Prof. C. S. Slichter to make the computations necessary to 
show in numerical terms what would be the reduction in surface area if 
the rotation were changed to the degree postulated in Darwin's interpre- 
tation of the past history of the earth and moon. It was thought by 
him sufficient to base the computations on the convenient hypothesis of a 
homogeneous density. The change of surface area was shown to be large 
and this made it clear that, if such a change of rotation has taken place, it 
is an important factor in deformation. Even if the chief deformation took 
place early in the history of the earth, the effects should be apparent still 
in the inheritances of the regions most affected, and the record should show 
them. For the purpose of a more critical study of the subject. Professor 
Slichter has been kind enough to recompute for me the requisite data on 
the basis of a distribution of internal density as near that of the actual 
earth as our present knowledge permits. For this purpose Laplace's law 

.» Van Hise, Jour. GeoL, vol. 6, pp. 10-64, 1898; Slichter, ibid., vol. 6, pp. 66-68, 1898. 


of increase of internal density was taken as perhaps the best expression 
of this factor and as being in fair accord with astronomical data. Pro- 
fessor Slichter extended the computation to other constants of the earth 
than those requisite for this inquiry and these give to his paper a value 
quite independent of its application to the present problem. His paper 
and table will be found on pages 61--67 of this volume. 

Column 7 of this table (page 67) shows that ten periods of rotation 
have been selected, ranging from 3.82 hours to the present period. Darwin's 
hypothesis * leaves unassigned the precise period of rotation when separa- 
tion took place, but from an inspection of the configuration of the spheroid 
at the rotation-period, 3.82 hours, and of the gravity in different parts of 
the spheroid at that stage, it seems safest to assume that a rotational 
period less than 3.82 hours would be necessary to cause fission. It seems 
best also to assume that at the 3.82-hour stage the earth was solid on the 
exterior, whatever may have been its internal condition. If this shall not 
seem so to any one, the arguments based upon the data of this rotational 
period can easily be shifted to the numerical values of the next period of 
4.03 hours, or to any of the later periods given in the table. 

From column 11 it will be seen that the equatorial circumference at 
the rotation-period, 3.82 hours, was 1,131 miles greater than it is at present, 
while the meridional circumference was 495 miles less. In changing to 
the present form, the tract immediately under the equator must have 
become shorter by 1,131 miles. The tracts under the parallels adjacent to 
the equator north and south would have become shorter by less amounts, 
those still farther away by still less amounts, until a little beyond 30^ 
latitude, north and south, parallels are reached under which the crust would 
have theoretically remained unchanged so far as this immediate factor is 
concerned. These are the latitudes of mean radius for each stage of rota- 
tion and are shown in column 9. It will be noted that these shift from 
lat. 33^ 20' to 35^ 13' in the course of the series, but it is sufficient for our 
purpose to speak of the neutral zone as lying at 35^ latitude, north and south 
respectively. The equatorial belt between these parallels, 70^ in width 
roundly speaking, would therefore, by the postulated change, have become 
shorter along its central line by 1,131 miles, since the rotation-period of 
3.82 hours. On its borders it would have suffered no change, and between 
the borders and the central line it would have suffered a graded series of 

That portion of the meridional circumference which lay within the 
equatorial belt should have been shortened in the course of the change 
from the rotation-period of 3.82 hours to that of the present, but the whole 
meridional circumference should have been lengthened 495 miles. It 
is obvious, therefore, that the areas north and south of the neutral zones 
must have become extended meridionally 495 miles plus the amount of the 
contraction in the equatorial zone, the precise value of which is unimpor- 
tant here. It will be convenient to call these areas of expansion polar caps, 
though they reach down to about 35^ latitude. In the course of the change 
named, the surface at the poles should have been raised and the curvature 

« The Tides, p. 360. 


of the c&ps iDcreased at all points. The extension should have been great- 
est at the poles and should have died away to aero at the parallel of no 
change. There was therefore a olimacterio stretching at the poles and a 
climacteno compression at the equator. 

Fig. 6 is intended to illustrate the nature of the ohange as seen from 
a point of view above one of the poles. To keep the view as true to pa- 
spective as practicable, the equatorial belt is foreehorteaed. The excess 
of area in the equatorial belt is represented by the black triangles, whioh 
are too small on account of this foreshortening. The deficiency of area in 
the polar repon is repreeented by the white ground, which ia more neariy 
in true proportion. 

Changes in the oruat of the earth of this magnitude, or of such leoso- 
magnitude as would have followed a change from any of the other eariy 
periods of rotation to the present one, could scarcely have taken place 
without leaving a record of them- 
selves in the form of compreesional 
and tensional phenomena. We 
may, to be sure, suppose that the 
interior of the earth has always 
been sufficiently mobile, in one 
form or another, to permit intei^ 
nal shift of material from areas of 
compression to areas of tension, 
and so to accommodate itself to the 
progressive change of form, but 
this can not reasonably be sup- 
posed to have taken place in the 
outer shell without having left evi- 
dences of itself, for this shell must 
be assumed to have been solid from 
an early state and, being at the 
surface, it was not under such prea- 
sure as to Sow and hence must 
have been deformed in the familiar 
modes that characterise surface 
thrust and tension respectively. It 
is known from abundant geological 
observation how the shell of the earth deports itself under conditions of 
compression and tension resulting from forces of the kind that would arise 
from the changes asogned. The data of Sliohter's computations may there- 
fore be interpreted by the usual methods. 

The equatorial belt of the earth of the 3.82-hour rotation-period would 
differ from that of the present earth to the extent of a broad swell 180 
miles high. In settling down this might doubtless relieve its excess of 
length in cross section by thrusting northward and southward into the 
areas of tension, but aa its equatorial length was 1,131 miles greater than 
the present equator, it would seem that in an east-west direction the tract 
must fold, crumple, and overthrust on itself after the familiar fashion of 



folded mountains. To estimate the result comparison may be made with 
estimates of the amount of crustal shortening involved in the formation 
of folded mountains. It is obvious that the estimates which assign the 
greatest amount of shortening to given amounts of folding are those which 
would give the least mountain production to the sinking of the equatorial 
belt in question and are hence the most conservative. One of the highest 
estimates of the crustal shortening involved in the formation of a familiar 
range of mountains, made by a competent geologist on the basis of much 
personal field work, is that of Professor Albrecht Heim for the formation 
of the Alps, which is 74 miles. Somewhat comparable estimates are those 
of Dr. Peter Lesley for the folds of the Appalachians west of Harrisburg, 
which is 40 miles, and that of Dr. G. M. Dawson for the Laramide Range 
in British Columbia, which is 26 miles. In the opinion of some other geol- 
ogists these estimates are too high. If therefore we apply these to the 
equatorial belt the results will be relatively conservative. If we use Heim's 
figure, the sinking of the equatorial belt to the assigned amount should 
give 15 mountain ranges of the magnitude of the Alps standing across the 
equator. They should be short ranges d3dng away within 35^ of latitude 
on either side. If we apply Lesley's estimate there should be 28 ranges of 
the order of the Pennsylvanian Appalachians standing across the equator; 
if the estimate of Dawson be used, there should be 45 ranges of the magni- 
tude of the Laramides of British Columbia. 

If we start with the 4.03-hour rotation-period instead of the 3.82-hour 
period, these figures become, 13, 26, and 40, respectively; and they may 
be easily reduced for later periods. 

If, as an alternative, we choose to assign mcnre mashing of the shell and 
less corrugation, it will merely give us a massive equatorial ridge with less 
cross-folding. If, as another alternative, we choose to assign more com- 
pression into denser rock, we shall have greater resistance to subsequent 
erosion and higher specific gravities to account for. 

Under no tenable hypothesis, so far as I can see, can an equatorial 
protuberance of 180 miles comparable to the 3.82'hour period, or of 180 
miles comparable to the 4.03 period, or of 87 miles comparable to the 6.35- 
hour period be assumed to have subsided to the present equatorial dimen- 
sions without having left a distinct record of itself in the form of transverse 
ranges of mountains, or of irregular protrusions, or of indurated terranes, 
or of some combination of these or of the other modes in which exceptional 
tangential stress is accustomed to express itself in the shell of the earth. 

It is to be noted that, by the terms of the retardational hypothesis, 
the tangential stress must have been applied constantly from the beginning 
to the present time. It was indeed more rapidly applied in the earlier 
stages, but some stress has been added constantly ever since. If compres- 
sion to a more compact form is to be assigned at all, in any important 
d^ree, it must be assigned to the first stages of stress, and the later pro- 
trusions would be all the more enduring on account of this early induration. 

Now it is for every one to examine for himself the equatorial tract to 
see if it presents the character which the hypothesis requires. For myself , 
I am quite unable to find it. There is not even an equatorial belt of land. 


much^less an elevated girdle accidented by cross-folds, or knotSi or con- 
torted protuberances; nor do I find evidences of the truncated remains of 
these. Since the rotation-period of 15.63 hours, 40 miles of shortening 
should have been added to all that preceded, and 15 miles of this should 
have been added since the 19.77-hour period. Even if these were remote 
in years, they should have served to perpetuate a phenomenon that in its 
nature must have been dominant from the beginning, for it is difficult to 
assign any other agency of deformation that should have overmastered 
this, if it had this degree of efficiency. On the contrary, other agencies of 
deformation should, according to an accepted generalization derived from 
observation, have reenforced the deformation assigned to this cause, for 
old lines of 3delding usually determine new ones. 

As a matter of fact the depressions below sea level on the line of the 
equator are fully as great as the amount normal to a great circle; about 
three-fourths of the equatorial zone is submerged and one-fourth emergent. 
The oceans crossed are normally deep; the mountains of the tract are 
scarcely normal in height or massiveness, the Andes of Ecuador being the 
only conspicuous range within the equatorial tract. The mountains which 
cross the equatorial tract show no special signs of limitation to it, as they 
should if they were essentially dependent on the agencies involved in the 
retardational hypothesis. 

If we take into consideration the whole compressional belt from 35® 
north to 35® south, it is found to embrace but little more than the average 
amount of land; indeed, the emergent surface within it is less, in proportion 
to the submerged area, than in the region north of it, though it is more 
than in the region south of it. 

If we turn to the tensional areas that should, under the hypothesis of 
reduced rotation, lie between 35® north and south and the poles, the inspec- 
tion is unembarrassed by any doubt about the effect of the stress upon 
the density of the rock, for appreciable stretching can not be assigned to 
rocks, except as it expresses itself by Assuring and equivalent modes, 
which leave an appropriate record. It is to be observed here again that, 
while the larger part of this tension was brought to bear in the early stages, 
it was, according to the hypothesis, continuous throughout the whole 
history. The results naturally assignable to this progressive tension would 
be a persistent fissuring and gaping radial from the poles, somewhat as 
implied by fig. 6. This must have run through all geological time, except as 
counteracted by some other agency. The cooling of the earth, or its shrink- 
age from internal molecular change or from any similar pervasive agency, 
would antagonize this, and if equal to it might prevent the actual opening 
of the fissures. But, to be consistent, this shrinkage must be applied gen- 
erally and such application would intensify the difficulties in the equatorial 
belt in proportion as it relieved those of the polar caps. Simply to counter- 
act the 495 miles of stretching required by the hypothesis in the rotational 
reduction from a period of 3.82 hours to the present, leaving out of con- 
sideration its special distribution, would require about 78 miles of vertical 
shrinkage in the polar regions and of 1,600 miles in the equatorial belt. 
But there is a special difficulty of distribution. The stretching required by 


the hypothesis of rotational reduction is concentrated toward the poles, 
and hence, if tension is to be avoided in high latitudes, a very much larger 
radial contraction than the amount named must be postulated. 

It must also be considered whether cooling, or any other similar con- 
tractional agency that can be postulated consistently with the early states 
of the earth assumed by this hypothesis, would be competent to offset the 
tensional effects imposed by the change of rotation in the polar regions. 

If, to escape the difficulties arising from exceptional tension in high 
latitudes, it be assumed that the whole shell of the lower latitudes crowded 
toward the poles, this would involve meridional crowding and the forma- 
tion of a system of folded ranges pointing to the poles, while east-and-west 
ranges should be absent proportionately, and thus the effects should be 
expressed in a distinctive manner. So it seems safe to conclude that, in 
one way or another, the high-latitude tension should have expressed itself in 
a characteristic way and, on account of its magnitude, its expression should 
be declared. 

In comparing the facts with the theoretical requirements it must again 
be noted that the earlier formations should show the most evidence of 
tension, the Archean most of all. As a matter of fact, the Archean of high 
latitudes, as of low latitudes, shows abounding evidences of compression. 
It was my privilege in 1894, as geologist to the Peary Auxiliary Expe- 
dition, to see something of the ancient crystalline rocloi of Greenland at 
latitudes as high as 77^. They bore the same evidences of crumpling, 
contortion, foliation, and thrust-stress generally as are commonly shown 
by the Archean rocks in lower latitudes. All descriptions of high-latitude 
formations of this age are identical in dynamic characters with those of 
lower latitudes, so far as my knowledge extends. The Archean terranes 
of Scandinavia and Finland lie far within the area of hypothetical tension, 
as do also those of Scotland and Canada, and even those of central Europe 
and the northern United States. The Archean and Proterozoic rocks of 
these regions bear evidences of tangential thrust of a most declared type, and 
no distinction between the most ancient rocks of the high-latitude and the 
low-latitude regions, in the matter of compressional characters, has, I believe, 
ever been detected. The literature of the subject does not show any special 
distribution of veins, dikes, normal faulting, and other evidences of tensional 
stresses correspondent to latitude. Apparently these]f eatures are essentially 
as prevalent in the equatorial belt as in the polar circles. 

If the equatorial belt has been subjected from the beginning to constant 
increments of tangential stress and of gravity (column 4 of Slichter's table, 
page 67) while the polar regions have been concurrently subjected to incre- 
ments oif tension and decrements of gravity (column 5), it would seem that 
volcanic action would always have found adverse conditions in the former 
region and favorable ones in the latter, certainly so if pressure is adverse 
to liquefaction and if tensional faulting facilitates eruption. It does not 
appear, however, that volcanoes are in any appreciable degree infrequent 
in the tropical zone or that they are specially frequent in high latitudes. 
The prevailing impression is that they are somewhat more abundant in 
the tropics than in high latitudes, but there is little, if any, warrant for any 
latitudinal discrimination. 


The distribution of folded mountains appears to be quite indifferent 
to the latitudinal distinction which the hypothesis of rotational reduction 
involves. The great Cordilleran belt of the Americas begins far within 
the southern tensional area, is strong where it crosses the southern neutral 
belt, is ako strong in the southern ha^ of the equatorial belt, becomes weak, 
scattered, and tortuous in the northern half of this belt, attains strength 
and broadens as it crosses the northern neutral aone, and reaches great 
breadth and aggregate mass in the lower part of the north tensional area. 
About 30^ within that zone, still strong, it swings about toward the Asian 
continent. The great tangled mass of mountains of central Asia lies chiefly 
in the northern tensional area. According to Suess, the thrust movement 
was generally from the northwest; that is, from the more highly tensional 
to the less highly tensional area. The great east-westerly range of southern 
Europe and Asia lies chiefly in the lower tensional and neutral zones and 
only at the east passes obliquely into the equatorial belt. A thrust from the 
compressional zone toward the tensional zone is indicated in the western 
portion and the opposite in the eastern portion. If, neglecting the latter, 
we fasten upon the former as dynamically probable under the hypothesis, 
it is to be noted that, with tension increasing in the direction of the thrust, 
it is not apparent whence came the resistance that was necessary to the 
intricate folding and distorting of the east-west ranges. Rather should 
we expect 3delding in the direction of the tensional area and lateral crowd- 
ing of the shell as it was pushed from the periphery toward the center of 
the tensional cap, with short meridisnal ranges as the result. Without 
reviewing the multitude of minor mountains, it may be sufficient to note 
that the Urals, the ranges of Scandinavia and of the British Isles, the 
Appalachians, and the mountains of Greenland testify to the dominance 
of thrust phenomena in the northern zone of tension. Statistically con- 
sidered, the facts now known give this northern zone precedence over all 
others in thrust phenomena. The great Archean tracts of Canada, Green- 
land, Scandinavia, and Finland carry the dominance of this thrust phenom- 
ena back to the earliest known ages. Taking the facts as we now know 
them, there seems to be no observational support for the compressional- 
tensional distribution which the hypothesis of great tidal retardation 

In the discussion thus far, agencies of compression and tension, other 
than rotational, have largely been ignored for the sake of following out, 
consecutively and iminterruptedly, the consequences of the hypothesis 
of rotational reduction and comparing them with observed facts. It is 
proper now to consider whether the intercurrence of other agencies of 
deformation would mask the results of tidal retardation, if these were of 
the order of magnitude implied by the fission theory of the origin of the 
moon, or even the close approximation of the moon to the earth in its 
early history under the planetesimal hypothesis. The existence of other 
causes of crustal deformation is of course fully recognized. To bring these 
under consideration in connection with the hypothetical tidal effects, it 
is necessary to note first their qualitative relations and second their relative 


Essentially all of the other assignable causes of deformation of the 
major class seem to be general in their application and to affect all latitudes 
practically alike. This is true of cooling, of internal redistribution of heat, 
of molecular rearrangement, whether chemical, crystalline, or diffusive, of 
atomic transformations and decompositions, of radioactivity and presum- 
ably of igneous extravasations. Grant to these agencies whatever sep- 
arate or combined effect may be their due, that effect, if it be general 
and essentially indifferent to latitude, as it seems that it must be, should 
be distinguishable from the effect of a superimposed agency that is pro- 
nouncedly correlated with latitude, because of this peculiarity. Granted 
a given amount of uniform earth shrinkage as the result of the general 
agencies named, or any of them, the crustal stress arising from this in the 
equatorial belt would be intensified by the addition of the stress of the same 
kind arising from the retardation of the earth's rotation, while the crustal 
stress which arises from these agencies in the polar regions would be propor- 
tionately relieved by the tension arising there from rotational retardation. 
A difference of result equal to the algebraic sum of the retardational and 
general stresses should be manifest in the resulting deformations. The 
conspicuousness of this difference must depend largely on the relative values 
of the two classes of agencies, which is our second point of consideration. 

If, on the tidal side, we take the higher deformative values given in 
Slichter's table (page 67), and if, on the other side, we take estimates of 
shrinkage made from a study of the foldings and faults of the earth, a 
comparison may be made. Quite without thought of this application, I have 
recently reviewed the data of the latter class in the endeavor to form a 
reasonable estimate of the amount of shrinking which the earth has prob- 
ably undergone; and, while this estimate has little claim to value in itself, 
it may perhaps be taken to fairly represent the import of the present 
imperfect data. It is as follows: 

If one is disposed to mimmize the amount of folding, the estimate may perhaps be 
put roundly at 50 miles, on an entire circumference, for each of the great mountain-making 
periods. If, on the other hand, one is disposed to give the estimates a generous figure so 
as to put explanations to the severest test, he may perhaps fairly place the shortening at 
100 n^es, or even more. For the whole diortening since Oambrian times, perhaps twice 
these amounts might suffice, for while there have been several mountain-making periods, 
only three are perhaps entitled to be put in the first order, that at the dose of Uie Faleo- 
soic, that at the dose of the Blesosoic, and that in the late Oenosoic. The shortening in 
the Fhyterosoic period was considerable, but is imperfectly known. The Archean rocks 
suffered great compression in their own times, and probably shared in that of all later 
periods, and if their shortening could be estimated dosely, it might be taken as covering 
the whole. Assuming the circumferential shortening to have been 50 miles diuing a given 
great mountain-folding period, the appropriate radial shrinkage is 8 miles. For the more 
generous estimate of 100 miles, it is Id miles. If these estimates be douUed for tJie whde 
of the F^eosoic and later eras, the radial shortening becomes 16 and 32 miles, respectiydy.' 

If we assign to the Proterosoic era a shrinkage equal to the Paleozoic, 
Mesozoic, and Cenozoic eras combined, and to the known Archean twice as 
much, the minimum and maximum estimates are 64 miles and 128 miles 
of radial shrinkage, respectively, or roundly 400 and 800 miles circumfer- 
ential shortening, respectively. 

' Geology, vol. 1, Chamberlin and Salisbury, 1904, p. 551. 


The assigned equatorial shrinkage from reduction of rotation since the 
3.82-hour rotation-period is 180 miles, which is to be compared with the 
minimum 64 miles or the maximum 128 miles of the above estimate. Of 
course, it must be recognised that the 180 miles covers a period preceding 
the known Archean, which is not embraced in the latter figures; but if an 
allowance of two-thirds be made for this, the remaining 60 miles vertical 
shrinkage still bears a sufficiently large ratio to the stratigraphicai estimates 
to make its effects certainly discernible, when the contrasted influences 
in polar and equatorial regions are brought into comparison. 

The computation for maximum rotational change gives a meridional 
elongation of 495 miles; the stratigraphic estimate gives a meridional 
contraction of 400 and 800 miles minimum and maximum respectively. 
Allowing two-thirds of the 495 miles for the period preceding the known 
Archean, there remain 165 miles of elongation to reduce the effects of the 
400 or 800 miles of contraction. 

Combining equatorial and polar effects, the case stands 777 (400+377) 
vs. 235 (400-165), on the minimum basis, and 1,177 (800+377) vs. 635 
(800- 165), on the maximum basis, when two-thirds of the retardation is 
assigned to pre-Archean times. It would seem that differences of this order 
of magnitude should be clearly manifest in the phenomena. 


If there be any doubt about the practicability of detecting the influ- 
ence of any great change in the rotation of the earth by the distinctive 
features of the deformation of its shell, we certainly have a very delicate 
means of detecting deformations in the position of the sea-level relative 
to the land. The position of the sea-level has been recorded by a series of 
shallow-water and shore deposits extending from the Cambrian period to 
the present, and this record was made with sufficient frequency and fidelity 
to answer every purpose of an inquiry of this kind. To a much greater 
extent than has usually been recognized, the known stratigraphic series 
is the product of shallow water, as shown by shallow-water life and appro- 
priate physical evidences. In many cases some latitude must be allowed 
in the interpretation of these criteria of depth, but this can be the source 
of no essential error in a problem of deformation whose units are miles 
rather than feet; but, if required, a sufficient number of cases of irreproach- 
able accuracy can be given, for at not a few geologic epochs there were 
emergences and submergences between which some stage of the transition 
marks the relations of the water surface to the land with positiveness and 
exactness. If, for instance, we know that in the critical regions, whether 
poleward or equatorward, a given horizon has been above the water- 
level and below the water-level respectively at two successive stages, we 
know that between these stages it was absolutely at the water-level. By 
means therefore of the successive emergences and submergences of given 
horizons, the relations of the sea to the land can be determined very accu- 
rately for a sufficient number of geological stages to be wholly decisive in 
such a problem as that in hand, and approximately for most of the other 


If the litbospbere could be supposed to have acted under tbe forces of 
gravity and rotation so nearly as tbough it were a perfect fluid tbat its form 
would be at all times perfectly adapted in all its parts to its rate of rota- 
tion, bowever mucb tbat may bave cbanged, the argument here introduced 
would bave little or no force. If this assumption is made here, it must 
of course be carried consistently throughout the whole range of deforma- 
tive interpretation. If this is done faithfully, very grave difficulties will 
be encountered, so grave that, for myself, I bave found them insuperable. 
It is indeed commonly thought consistent with experiments and geological 
observations, to regard tbe litbospbere as a solid which acts rigidly toward 
stresses of short period, and quasi-fluidly towards those of long period. 
Under this proposition it is possible to assume tbat the accommodation of 
the earth to a steady change of rotation might be so nearly perfect tbat 
▼ariatioiis would escape detection by even so delicate a registration as 
that of tbe sea-surface. But if this is done, it should be with tbe full con- 
fldousness tbat this is not a deduction from the proposition, but merely 
an assumption under it; for the general proposition that tbe litbospbere 
will yield under stress applied for a sufficient time does not in itself carry 
tbe conclusion tbat it will yield under the given stress in tbe given time. 
A quartz crystal is under self-gravitative stress and may bave also been 
under terrestrial gravitative stress for eons, and yet it shows no signs of 
becoming a gravitative spheroid. Mountains and continents are under 
gravitative stresses and they probably jidd to these, but at wtiat rate is a 
practical question of much geological importance. Tbe postulate of quasi- 
fiuidal accommodation is not a solution; it is only a broad generalization 
under which a solution may be sought by specific evidence. 

The shell of tbe earth is chiefly an aggregate of interlocking crystals 
which are possessed of specific elasticities of form, and the whole aggregate 
dearly has elasticity of form. If the great mass of the earth or even the 
deep outer portion be similarly an elastico-rigid solid, deformations will 
only take place when the stress-differences rise to equality with the elastic 
resistances, except in the limited form of strain, and to tbe limited degree 
permitted by tbe individual transfer of molecules from one rigid attach- 
ment to another. Deformations in this case await a certain accumula- 
tion of stress-difference. As the crux of the whole deformative problem 
lies largely in these basal conceptions, we may do well to turn to geological 
phenomena to ascertain, if possible, whether the earth does habitually 
yield concurrently with the accumulating stress-differences and thus con- 
stantly accommodate itself to stress-demands, or whether stress-differences 
do actually accumulate until the elastic limit is reached when deformation 
proceeds with relative rapidity until an approximate equilibrium is reached. 
This is but stating in dynamic terms tbe question of periodicity in geo- 
logical deformation. On this question, a consensus of geological opinion 
can not now be cited without qualification. Apparently views differ and 
reserve predominates among cautious geologists. It appears to me, how- 
ever, that strong evidence is steadily accumulating, from various quarters 
of the globe, tbat there were great periods of base-leveling of essentially 
world-wide prevalence, with concurrent sea-transgression, separated by 


briefer periods of deformation of similar prevalence. For these it does not 
appear that there will be found a consistent explanation except in the 
ability of the great body of the earth to accumulate stresses to a notable 
degree during the long periods of relative quiescence necessary for the base- 
leveling and the sea-transgression. 

It would be going beyond the proper limit of this paper to try to estab- 
lish this thesis by the citation of evidence, for this would involve a review 
of some large part of the great mass of stratigraphic, paleontologic, orograic, 
and physiographic data possessed by geology. Suffice it therefore to note 
here that this is the one of the alternative views of the earth's deformative 
methods that seems at present best supported by geologic evidence. It 
is not wholly necessary to the following considerations, though it lends much 
strength to them. 

Let it be assumed merely that the earth-body ofifers some appreciable 
resistance to deformation, an assumption which can scarcely be questioned, 
since the irregidarities in the form of the geoid imply this, even when allow- 
ances are made for differences in the distribution of density. Let a limited 
slackening of the earth's rotation take place. This will disturb the preced- 
ing equilibrium between the centripetal and centrifugal forces and both the 
body of the earth and the water on its surface will experience stress-differ- 
ences which give a tendency toward a new equilibrium. This equilibrium 
may be established by the subtraction of matter from the equatorial regions 
and its transfer to the polar regions internally or externally. The earth- 
body certainly offers some resistance to this transfer while the water on its 
surface offers practically no resistance at all because it is in circulation as 
the result of solar influence, and to effect the new distribution it is only 
necessary that it stop where the new demands of gravity require, and in 
this friction will lend its aid. The water surface may therefore be supposed 
to fall in the equatorial regions and rise in the polar regions until the new 
water surface of the globe conforms to the new equilibrium required. 
This must relieve, in some large part at least, the stress upon the body of 
the earth, for if the newly developed equilibrium required more matter 
in the polar regions the water would supply it, unless it were previously 
exhausted. Local stresses might remain where the land was left pro- 
tuberant, but geological evidence shows that such protuberances can be 
maintained for long periods by the effective rigidity of the earth, if they do 
not exceed a certain measure. Such a protuberance of the equatorial land 
may be treated as any other local protrusion of the earth's body. When, 
therefore, in the case in hand, an equatorial mass became protuberant 
above the surface of the geoid sufficiently to overcome the effective rigidity 
of the part of the earth affected, the appropriate deformation would follow. 

The determination of what mass is sufficient for such deformation is 
qualified by the available time. Given infinite time and the requisite mass 
would doubtless be relatively small in a body like the earth, even on the 
hypothesis of elastic rigidity; for, even within the limit of elasticity, 
deformations may take place by the transfer of molecules from one rigid 
attachment to another individudly. But whatever might be the results if 
indefinite time were available, the practical case is one of limited time and, 


as implied above, geological evidences seem to show that stresses do accu- 
mulate to certain large magnitudes before sensible deformations take place. 
Meanwhile surface transfers by wind and water action are in progress. 
The protuberant equatorial belt postulated must ever have been shedding 
material northward and southward, mechanically and by solution, thus 
building up sedimentary series in the flanking sea-borders. As the pro- 
tuberant tendency was ever renewed by slackening rotation, this should 
have become a perpetual process and, as we have seen, should have been 
a pronounced factor, if not the dominant one, in the earth's deformation, 
if the reduction in rotation was as great as the hypothesis of earth-moon 
fission requires. It appears, therefore, that an annular latitudinal dis- 
tribution of the sediments and of the lands derived from the sediments 
should have arisen, and this should have codperated with the tendency of 
the waters to polar accumulation in giving a distinctive configuration to the 
distribution of land and water. Yet, as a matter of fact, the surface con- 
figuration is singularly free from latitudinal sones. There is a very rough 
tendency toward a meridional arrangement, but the essential fact is that 
the arrangement is irregular. The protuberances and depressions consist 
of an unsymmetrical interspersion of independent triangidar, quadrangidar, 
oval, and scarcely definable areas. 

Going more into detail, and in this insisting only on the obvious general 
proposition that the water-level in the equatorial sone should have tended 
to a low position relative to the land and to a high position in the polar 
regions, we may note that the Greenland Archean embossment not only 
stands high above the water-level to-day but is singidarly free from evi- 
dences of submergence in the past. At various periods from the Cambrian 
onwards, the water-level has stood low about its base and has risen above 
and fallen below the present shore-line. Much the same may be said of 
the great Archean tract of Labrador and of the region west of Hudson's 
Bay, as also of that of Scandinavia and Finland. It is a remarkable fact, 
in the light of the matter in hand, that the old lands which are now best 
exposed, the lands that seem to have been longest out of water, and that 
have been most persistently above sea-level, are more largely the lands of 
high latitude than of low latitude. 

A candid and critical survey of the relations of land and water in high 
and low latitudes alike, and in all longitudes, especially in the northern 
hemisphere where best known, and where the protuberant lands furnish 
the best record, seems to me to reveal a singidar constancy of relations, 
subject only to oscillations measured by a few thousands of feet at most, 
an order of magnitude quite out of harmony with any hjrpothesis which, 
to dte a very conservative example, requires that the equatorial tract 
should have been 8 miles higher than at present when the rotation-period 
of 14 hours prevailed. 

If the moon were once much nearer the earth than now the tides should 
have been much stronger and the littoral deposits of the early ages should 
show not only greater coarseness but greater vertical range. Geologists 
have not been generally convinced that the earlier sediments are different 
in any such systematic way from those of later times. 



If the earth's rotation were much more rapid than now in early times, 
the gyratory component affecting the courses of the winds would have been 
strengthened and probably trees would have required a corresponding 
strengthening of the trunks, branches, and roots to meet this successfully. 
Such provisions are not certainly detectable. In the coal-accumulating 
eras trees grew to great heights without tap roots, and in some cases they 
appear to have grown on accumulations of vegetal debris which could not 
have furnished a very secure hold, and yet there is no evidence that they 
were especially subject to overthrow. In no way is it clear that the life of 
the early ages, either vegetal or animal, was adapted to atmospheric move- 
ments essentially different from those of to-day. 

A more rapid rotation should have caused a stronger deflection of the 
streams to the right hand in the northern hemisphere and to the left in the 
southern. This should have resulted in tilted aggradation planes. How- 
ever, these might not now be capable of detection, even if present. 

It is probable that some changes would arise from the shortness of the 
day and night, but it is not clear just what these would be nor what would 
be the criteria for their detection. 

It seems safe to say, in summation, that no geological evidence of any 
unquestionable kind, or even probable kind, is found that supports the 
theoretical postulate of a former high rate of rotation of the earth. 

The geological criteria are not delicate enough, however, to forbid the 
belief that the rotation of the earth has changed in some minor degree 
during the time over which the record extends. If the deformative effects 
of such changes were small compared with those of the other diastrophic 
agencies, they might be so far^masked as to escape ready detection. 


There are some agencies, apparently not very potent ones, which tend 
to accelerate the earth's rotation and to offset the influences of the tides. 
Of these the most familiar is the shrinking of the earth. It was noted in 
the review of the hypothesis of Darwin that in the initial stage the shrink- 
age of the earth was made more effective rotationally than the tides of the 
sun. It was of course assumed as a basis for this that the loss of heat at 
that stage was quite exceptionally great. The computations of Wood- 
ward ^ and others have shown that the present rotational effects of loss of 
heat, assuming the correctness of current estimates, is exceedingly small. 
Even if the estimates of loss of heat need to be increased, as seems probable, 
such loss can not be a very efficient agency. Shrinkage from other sources, 
as molecular rearrangement, atomic reconstruction, or other agencies, may 
have a more considerable effect. The rotational results of the contraction 
of the body of the earth from a radius of 4,160 miles to 3,960 miles, with 
intervening stages, as computed by Dr. MacMillan on the assumption that 
the Laplacian law of density is maintained, are as follows: 

^ " The effects of secular cooling and meteoric dust on the length of the terrestrial 
day." < Astro. Joiir., No. 502, 1901 : ^' From this it appears safe to oondude that the length 
of the day will not chanse, or has not changed, as tne case may be, by so much as a half 
second in the first ten mulion years after the initial epoch. ** p. 174. 



LmngOi of day for variouB lengths of earih's raditu — Laplacian law of dennty. 

Radiu of earth 


Length of day. 


24h 0^ 0* 



24 7 17 



24 14 36 



24 21 64 



24 29 14 



24 36 36 



24 43 68 



24 68 46 



26 13 39 



26 61 10 



26 29 08 

If current views of the shrinkage of the lithosphere founded on the folds 
of the shell and on overthrust faulting are valid, its accelerative effects on 
the earth's rotation are greater than the retardative effects of its water-tides 
as hereinbefore computed, and perhaps greater than all the tides combined. 

It is possible that the earth may respond to the radiant energy of the 
sun as a thermal engine and that its rotation may be influenced by this, 
but the subject is obscure and elusive, if not delusive, and no attempt will 
be made to develop it here. 

If, as seems probable, the evening sky, because of clouds, dust, etc.» 
offers more resistance than the morning sky to the passage of solar radiation 
tangential to the earth, there is a slight preponderance of light-pressure in 
favor of acceleration of rotation, but it must be very small. 


The application of the most radical and the most rigorous method of 
estimating the frictional value of the present water-tides, a method which 
brings to bear practically all the friction of these tides as a retardative 
agency, irrespective of their positions or directions of motion, seems to 
show that they have only a negligible effect on the earth's rotation. 

From the best available evidence I conclude that the tides of the litho- 
sphere are chiefly elastic strains and have little retardative value, while 
Uie tides of the atmosphere are too small to be measured. 

The accelerative influences seem to be also negligible, so far as geological 
applications are concerned. 

In close accord with these deductions, the geological evidences indicate 
that there has been no such change in the rate of the earth's rotation 
during its known history as to require it to be seriously considered in the 
study of the earth's deformations. 

I desire to acknowledge my great obligations to my colleagues, C. S. 
Slichter, F. R. Moulton, A. G. Lunn, and W. D. MacMillan, for the indispen- 
sable aid which their several contributions have rendered to these studies 
and for criticisms and suggestions relative to my own paper. 





Profe$9or of Applied MathemaHa, UnivertUy of Wiicontin 




It is a simple problem to determine the rotation-period of an ellipsoid 
of revolution^ if it be postulated that the density of the body is uniform, 
and that the form is that assumed by a perfect liquid under like conditions 
of rotation. A table of the rotation-periods of such a body having the same 
volume and mean density as that of the earth, computed for various values 
of the eccentricity of an elliptic meridian section, will be found on p. 327 of 
Part II of Thomson and Tait's Natural Philosophy (edition of 1890). It 
is the purpose of the present investigation to obtain analogous results for 
an ellipsoid of variable density, assuming a law of increasing density from 
surface to center approximate to that actually possessed by the present 
earth. The law of density assumed in the computation is the well-known 
law of Laplace: ^^ sing a ^^^ 

in which the symbols have the meaning given on page 64. According 
to this law the internal layers or shells of equal density gradually change 
from the shape of the surface to forms more and more nearly spherical 
as the center of the spheroid is approached. The forms of these layers 
are best expressed in the case of ellipsoids of revolution by the ellipticity 
of a meridional section. This number is computed by subtracting the 
length of the polar or short axis from the length of the equatorial or long 
axis and dividing the result by the length of the equatorial axis.^ The 
variation in the value of the ellipticities is shown by the dotted line in 
fig. 7. In this diagram the polar axis is represented as divided into ten 
equal parts. The ellipticities of the shells of equal density are expressed 
as percentages of the ellipticity of the surface. Thus the ellipticity of the 
shell that cuts the polar axis at 0.5 of the distance from the center to the 
surface is equal to 85 per cent of the surface ellipticity, while the ellipticity 
of the central shell is about 80 per cent of the surface ellipticity. 

If we assume a surface density of 2.75 and a mean density of 5.50, the 
above expression takes the form: 

4.365 ao . 2.4605 a ,^, 

/o = ^sm (2) 

The variation in density according to this law is shown graphically by 
the continuous curve of fig. 7. 

' The mean axis or the polar axis is often used as the divisor. There is little differenoe 
in the^results for the small values of the ellipticity usually involved. 




An ioBpeotion of the diagram sbowB that the dennty inoreuM quite 
uniformly tor a considerablfl distance as we pan from the Barfaee tovard 
the center. We finally come to a central nueleua oS naariy uniform deooi^. 
The density at the crater, required by the Lapladan law, is 10.74. ^s 
value would be modified if values different from 2.75 and S.SO be aanuned 
for the surface dendty and mean density respectively. 

1 1 1 1 1 I 1 1 1 X li 


Polar Axis of Spuaoio 


a« ■■ mean radius of surface of spheroid. 

a -= mean radius of any homogeneous shell in interior of spheroid. 

6 -= equatorial radius of any homogeneous internal shell of sphwoid 

of revolution. 
&s " equatorial radius of surface of spheroid. 

c^ polar radius of any internal homogeneous shell of spheroid. 
Co» polar radius of surface of spheroid. 
0-=a constant '=4.366 for earth-epheroid. 

q = & constant B' 2.4606 for earth-spheroid. 

/> = density of any homogeneous shell of mean radius a. 
/)q — surface density of spheroid - 2.76. 

a^ — equatorial attraction or value of gravity at equator of any spheroid. 
Qp— polar attraction or value of gravity at pole of any spheroid, 
m — ratio of centrifugal force at equator to gravity at equator. 

e —eccentricity of any homogeneous internal shell of spheroid. 

fg ~ eccentricity of surface of spheroid. 



-r— — ellipticity of any homogeneous internal shell of spheroid. 

c, " ellipticity of surface of spheroid. 
M » centripetal acceleration at equator of spheroid. 

While the Lapladan law of density originated in an assumption of 
Laplace which had little to recommend it beyond mere plausibility, the 
law is now believed to be fairly close to the truth. The computed values 
of the earth's precession based upon this law of density agree well with 
the observed values. The law is probably quite as near to the truth as is 
the measured value of the earth's mean density, which must enter as a 
basal number into any formula of density we may adopt. 

The plan of the investigation is substantially as follows: The attrac- 
tion of the heterogeneous spheroid of given ellipticity is first found for 
points on the equator and at the poles of the spheroid. These results are 
substituted in Clairaut's well-known equation connecting gravity at the 
pole and at the equator with the equatorial centripetal acceleration, and 
hence with the rotation-period of the earth. In thii9 manner the rotation- 
period for any given ellipticity of meridional section becomes known. 

Clairaut himself gave expression to formulas which give the attraction 
at external points of any rotating liquid ellipsoid.^ For polar and equa- 
torial points these may be written as follows (referring to the preceding list 
for the meaning of the symbols) : 

Equatorial ( f<^o f^t \ 

attraction = a, —|.-^,KJ / pda^+1..^ j pd(flh)l (3) 

attraction = 

In both formulas £ is a constant whose value depends upon the units 
of measure in which the various magnitudes are expressed. 

These expressions assume that each stratum of density p has the 
ellipticity t that would exist for the given rotation-period if all strata 
were perfectly liquid. In other words, the formula is built upon the hypoth- 
esis of the perfect fluidity of the spheroid. If we apply these expressions 
to the present earth we assume that the rigidity of the earth' is not sufiScient 
to withstand for geological intervals of time the stresses that would exist 
if the form of its surface differed materially from that of a free liquid. 

The above expressions (3) and (4) can not be integrated until we sub- 
stitute for p the appropriate law of density from (1) above. Using the 

<p(a)=-^ (sin qa^—qa^ cos qa^) (5) 

> See '' History of the theones of attraction and the figure of the earth/' l^ I. Todhi^ 
London, 1873, vol. 1, p. 220. 


we may obtain 

'pd{ah)^5a.\i.^^)<p{a.) / pda*^df{a.) (6), (7) 

from which we obtain 

..-^K.W(l+f[^-f]) (8) 

..-ft.W(i-^[.-f]) (.) 

The relation between m and c^ depends upon the ''degree of hetero- 
geneity" of the earth: that is, it depends on the departure of the surface 
and central density from the mean density. For a homogeneous earth we 
may write, as is well known: 

2c.-|m (10) 

but for a body possessing the law of heterogeneity given by equation (2) 
above, we can deduce the expression ^ 

The change of the denominator of the fraction on the right-hand side 
of the equation from 2 to 2.536 is brought about by the change of hy- 
pothesis from 

{Central density™ 5.50 
Mean density » 5.50 
Surface density — 5.50 
to the hypothesis 

{Central density » 10.74 
Mean density «- 5.50 
Surface density — 2.75 

If we further assume that the mean attraction at the surface of the 
earth is 982 dynes per gram of attracted matter, we may write the equa- 
tions of equatorial and polar attractions in the simple form 

a. = 982(l-0.1739eo) (12) 

ap = 982(l+0.3477eo) (13) 

M = 982(1.0144)eo (14) 

It is upon the numerical values here written that the results of the 
following table have been obtained. The results can be checked by sub- 
stituting for e in equations (3) and (4) the expression 

£ = [0.09645 (a»+a») +0.8071]€o (15) 

which is an algebraic function approximately equivalent to the transcen- 
dental relation between e and a, within the interval with which we are 

* See a "Treatise on attractions, Laplace's functions, and the figure of the earth/' by 
John H. Pratt, London, 1871, p. 116. 



concerned. This expression was found in 1898 by Mr. H. C. Wolff, then 
a graduate student at the University of Wisconsin. 

Investigations concerning the properties of the spheroid have usually 
hjrpothecated an ellipsoid so nearly spherical that small error would be 
introduced by neglecting the square of the ellipticity in comparison with 
its first power. Such has been done in the present instance. For that 
reason it is hardly possible to extend the computations to spheroids of 
greater ellipticity than those given in the table. As a matter of fact, the 
writer believes that he has extended the computations as far backward as 
is practicable without straining the approximate formulas beyond their 
limit of significance. 

The Rotation-period of a Heterogeneous Spheroid, 


Polar radius 

(mean radius 

= unity). 





(mean radius 

= unity). 













at equator 

(dynes) . 


period of 



period of 

















































• • • • 

• • • • 



• • • . 

• • • • 
• . . . 


Latitude of mean 


Equatorial contrac- 
(per cent). 


Equatorial contrac- 


Meridional elonga- 
(per cent). 


Meridional elonga- 

35* 13" 
35 12 
35 10 
35 8 
34 47 
34 35 
34 27 
34 15 
34 5 
33 53 
33 45 
33 33 
33 20 












































(mTRntunoNS to cosHoeomr and the fundamental problems op geology 







In this paper the waters of the ocean will be conceived as concentrated 
in a basin, rectangular in shape, the width of which will be taken as 2,860 
miles and the length as 60,000 miles. The bottom will have a uniform 
slope from the surface to a depth of 600 feet at 100 miles from the shore, 
dropping then to a depth of 9,000 feet at a distance of 150 miles from the 
shore, and then parallel to the surface out to the middle of the basin, the 
opposite side having the same shape. The tide will be supposed to rise 
4iV f^t in 6 hours — falling at the same rate. 

Sec. a I Sec.b 

I so mi. 

Sec. c ! 

1280 jni. \ 


Fio. 8. CroM-teoiion of barin (ahowing oim-IiaII). 

The rigorous determination of the motion of the water in such a basin 
on the principles of hydrodynamics seems to be unattainable at the pres- 
ent time. It is true that^ to start with, we have the equations of motion of 
a viscous incompressible liquid, but I have not succeeded in finding a solu* 
tion for them with the assigned boundary conditions, and therefore am not 
able to give an exact statement of the rate at which energy is dissipated. 
We may, however, approach the problem through some of the formuka 
of hydraulics and obtain an approximation, which, even though it be 
rough, will permit us to form some idea of the order of magnitude of dissi- 
pation. If we liken the ebb and flow of the tide to the flow of water in a 
canal we can use the formula of engineers for the loss of head due to fric* 
tion and viscosity, and consequently the loss of energy. 

Weisbach * gives us the following formula: 

A-^vl irih V ^^^^^ vel.y wetted perimeter 
"" ^^ 2g area of cross-section 

where h is the total fall of water in the canal necessary to maintain the flow, 
( is the coefficient of friction, and g is the acceleration of gravity. We will 

* ThecMretical Mechanics. Translated by E. B. Coxe. 8th Amer. ed., sec. 475. 



take (jf » 32 feet per second per seoond. The coeflBcient $ itself is dependent 
upon the velocity. As a result of many experiments the following value is 


.1920 \ 
V ) 

where v is the mean velocity. From this formula it will be observed that 
$ increases as the velocity decreases. 

This formula is applicable to a canal in which the cross-section is uni- 
form throughout its length. In order to adapt it to a canal of variable 
cross-section and velocity an integration is necessary. Consider an element 
of the canal between parallel cross-sections at distances I and Z+cK from the 
upper end, and put 

V » velocity a «> area of crossHsection p » wetted perimeter 

We have then, from the above formula, 

dl "^2»a 
or, since 2^«64 and 

f = .007409(1+^) 
this may be written 

This expression represents the slope at the point I necessary to main- 
tain the flow. The rate of fall of the water in this element of the canal is 
obtained by multipljring the slope by the velocity, that is 

r-3f--«rate of fall of the water 

The distance through which the water falls multiplied by its weight gives 
the amount of work done expressed in foot-pounds. Consequently 


where E is the amount of work done per unit time and w is the weight of 
water in the element considered. The volume of water is equal to adl and 
the weight of a cubic foot of water is approximately 62.3 pounds. Therefore 

11; -62.3 adl 

These values substituted in the expression for E give 


g ^ 62.3X^7409 ^ (v* + A92W)pdl 


Consider now a section of the basin 1 foot wide, and by means of this 
formula compute the loss of energy in this strip in one second of time with 
such a tide as has been supposed. 


Let us begin with section '' a. '' We will suppose the surface of the water 
is level throughout the section. A tide of 4^^ f^t in 6 hours means a 
constant flow throughout the section of 600 feet per hour or one-sixth 
foot per second, i.e., v^l. The wetted perimeter is the bottom only, 
that is p»l, and thiis also is constant throughout the section. We have, 



„ 62.3X0.007409 / F/l V^n mon/l Vlwi 

where E^ is the work done in section ''a'', the length of the strip being 
528,000 feet. Evaluating the above expression there results 

E„ a 37.939 foot-pounds per second 


The flow in section ''a" shows that the surface sinks at the rate of 
j^ foot per second. We will suppose that the surface of section '^b*' 
sinks at the same rate, remaining always level. If Z be measured from the 

beginning of the section the velocity at the distance I is given hj v^^, 

when q is the volume of water flowing by the point I and a is the area of 
the cross-section. Since the surface sinks j^ foot per second 

?-100+^ a-600+ " 

5280 220 


1 528000+^ 

*" 6 ^528000 +2« 

The wetted perimeter is constant throughout the strip, so that p»l. 
The length of the strip is 50 mileB, or 264,000 feet. The dissipation of 
energy in this strip is then 

„ 62.3X0.007409 
E, p 

ni r 528000+f y 0.1920r 528000+f V \„ 
6»L628000+2aJ "*■ 6* L528000+2«J / 

By putting l~2MOO0x this expression becomes 

„ 62.3 X 0.007409 X 264000 
*" 64X216 



which may abo be written 

» 62.3X0.007409X264000 , >it ■ *' i . oooeai, . ^' i i^ 
^*" 512X216X14* * ^ll + ^-T-r7-l+32.256|l + ^^v^| V(fa 


From the reduction of this expression is obtained 

£5 »i 1 . 1 24 foot-pounds per second. 


Making the same assumptions with respect to this section that were 

made with respect to section "b", we find in the formula v^^ 


'-^^o+sAo °-»*^ 

80 that 


9000 X 5280 

As before, we take p »i 1 . The length of the section is 1 ,280 miles, or 1,280 X 
5,280 feet. If now we put {«5280xl280x, the expression for the lost 
energy is 

^ _ 62.3X0.00m9X^280X 1280 f |(i5 + i28»)«+172.8 (16+128.)*}d« 

which reduces to 

E^ » 142.484 foot-pounds per second 

Combining these results, we have 

£;«-37.939 S^-1.124 5,-142.484 Total, 181.547 

The total work done in the entire strip is therefore 181.547 foot-pounds 
per second. If the work done on the opposite shore of the basin be the 
same, the total work done per linear foot of basin is 363.094 foot-pounds 
per second. In a basin 60,000 miles long this would amount to 11,503 X 10^ 
foot-pounds per second, or 36,300 X 10^^ foot-pounds per year of 365} days. 
The kinetic energy of the earth due to its rotation is given by the 
formula ^^^j^ 

where T is the kinetic energy, / is the moment of inertia of the earth, and 
a> is its angular velocity. The moment of inertia depends upon the law of 
density of the earth's interior which is not known. We will probably be 
not far astray in using the Laplacian law of density, t.e., 

sm m— 

5 = ^ 


* See Tisserand, M4c. C^., 2, p. 234. 


where S is the density at the distance r from the center, a is the earth's 
radius, and 

m - 141^ 40' 28'^ = 2.4727 - 4.426 

With the Laplacian law of density, then 


/ iih^'^i^iii) 

—•=-;ra'— i[(3m'— 6) sin m— (m*— 6m) cos ml 
o m 

With the same law of density the mass of the earth (Af ) is 


'jf 7'^''"'i^(i) 

— 4;ro*— iFsin m^m cos ml 

Consequently, by the division of these two expressions, 

.^ 2 r(3m'— 6) sin m— (m'— 6m) cos ^ Iw 2 
""am^L sin m—m cos m J 

Substituting in this the numerical value of m, there results 

and therefore 


Taking the radius of the earth at 3,968 miles, its mean density at 6.5, and 
the sidereal day as 86,164 seconds, it becomes 

T - 159 X 10" foot-pounds 

At the rate of loss due to tidal friction as calculated above, this amount 
of energy would last 43,900X10* years. The day would be lengthened 
by 1 second in about 500,000 years. 





Aiiociate Profenor of A$tronomy, UnivenUy of Chiooffo 






Sir George Darwin has written a number of classical memoirs on the 
subject of tidal friction which are remarkable not only for the profundity 
and the thoroughness of the mathematical anal3rsis, but also for the charm 
and lucidity of the exposition of the nature of the problems treated, the 
hypotheses upon which the investigations were based, and the conclusions 
which were reached. Frequent references will be made to these memoirs 
in this paper, and for simplicity they will be designated by numbers as 

1. On the bodily tides of viscous and sezni-elAstio spheroids, and on the ocean tides upon a 

jrielding nucleus. <Phil. Trans, of the Royal Soo., Part 1, 1879, pp. 1-^. 

2. On the precession of a viscous spheroid, and on the remote history of the earth. <Phil. 

Trans., Pfeirt H, 1879, pp. 447-538. 

3. On the secular changes in the' elements of the orbit of a satellite revolving about a 

tidally distorted planet. '<Ph]l. Trans., Plirt U, 1880, pp. 713-891. 

4. On the tidiJ fdotion of a planet attended by several satellites, and on the evolution of 

the solar system. <Phil. Trans., Part U, 1881, pp. 491-535. 

5. The determination of the secular effects of tidal friction by a graphical method. < Pro- 

ceedings of the Royal Society of London, vol. 29 (1879), pp. 168-181. 

Darwin's method of treatment is to express the tide-generating poten- 
tial as a sum of terms, each of which is the product of a second-order 
solid harmonic and a simple time harmonic, and then to derive the corre- 
sponding surface harmonics which define the tidal deformations when the 
system has assumed a condition of steady movement. The results are 
adapted to viscous or elastico-viscous spheroids, the heights and lags of 
the several tides being expressed in terms of the speeds of the tides and 
the viscosity, or the rate of decay of elasticity of the tidally distorted body. 
The effects of these tides upon the motions of the disturbed and disturbing 
bodies are then derived with rare skill. 

Apparently Darwin's work can be questioned, if at all, only where he 
applies his analysis to the earth-moon system. Here he reaches the con- 
clusion that very probably the moon once separated from the earth by 
fission, and that it has been driven to its present distance by tidal friction. 
In reading these conclusions we should heed his warning:^ 

TIm lesolt at which I now arrive affords a warning that every conclusion must always 
be read along with the postulates on which it is based. 

^ 2, p. 532, footnote. 



It is quite evident from Darwin's discussions that he acceptSi as a general 
basis for reasoning on the problems of cosmogony, the Laplacian nebular 
hypothesis with its implication of a one-time fluid earth; indeed, in 2, 
p. 536, line 3, and in 4, p. 530, first paragraph, he explicitly states that 
he adopts this hypothesis in its main outlines. It is obvious that this 
point of view might tend to give one a confidence, perhaps without his 
realizing fully the postulates upon which it was based, that, even if the 
spherical harmonic analysis should not be strictly applicable to a hetero- 
geneous earth whose liquid parts are broken up by continental masses, 
it still would be sensibly correct when applied to a fluid body such as the 
earth was supposed, according to this theory, to have been in the past. 
It is now known that there are very grave, and I believe fatal, objections 
to the Laplacian ring theory. At any rate, one would not now make it a 
postulate in a discussion involving so many and such serious complexities 
as arise in the theory of tidal evolution, or allow it seriously to influence 
his conclusion as to what is the most probable of the various possible 
hypotheses. Darwin examined with great thoroughness the character of 
the results for various conditions of viscosity and semi-elasticity, and only 
where he undertook to say what seemed to him the most probable of vari- 
ous possible series of events was he influenced, possibly, by his preconcep- 
tions as to the early condition of the earth. To illustrate the delicacy of 
the discussion we shall enumerate a few of his conclusions together with the 
hypotheses upon which they were based. 

In 3, Part IV, and in its summary, pp. 871-876, Darwin discussed 
the inclination of the moon's orbit and the obliquity of the ecliptic Con- 
sidering first the hypothesis of imaU viscosity and tracing back the qrstem 
until the day and month were equal, he found that, if this hypothesis is 
true, the lunar orbit and the earth's equator must initially have had con- 
siderable mutual inclination. " If this were necessarily the case, it would 
be difficult to believe that the moon is a portion of the primeval planet 
detached by rapid rotation, or by other causes. " (3, p. 873.) Then taking 
up the hypothesis of large viscosity and supposing that it was 'Marge 
enough," he found, tracing the system back, that when the day and month 
were of equal length, then the lunar orbit was sensibly in the plane of the 
earth's equator, which was inclined 11^ or 12^ to the plane of the ecb'ptic. 
His final conclusion from this discussion (pp. 875-876) was that it wfll be 
most nearly correct to suppose that the earth in the earliest times, though 
plastic, possessed a high degree of stiffneaa, and that now the greater part if 
not the whole of tidal friction is due to oceanic tides, and not to bodily tides, 
for in this way the theory of the fission of the parent mass into two bodies 
and the present inclination can be best reconciled. 

In 3, Parts V and VI, the effects of tidal friction upon the eccentricity 
of the lunar orbit were considered. The equations were integrated on the 
hypothesis of small viscosity, and it was found that in past times the 
eccentricity was much smaller than at present, nearly vanishing when the 
day and month were equal. If it had been assumed that the viscosity was 
very large, the eccentricity of the lunar orbit would have been the greater 
the farther back the system was traced. Since a large original eccentricity 


is incompatible with the fission hypothesis, it is found necessary to 
conclude that, at least for a part of the earth-moon history, the viscosity 
of the part of the earth distorted by tides has been amaU. Hence that 
viscosity which best explains the inclination of the lunar orbit causes 
trouble when considering its eccentricity. Since numerical details were 
not worked out in the discussion of the eccentricity for large viscosities, it 
is not known to what quantitative extent the two things are antagonistic, 
and it may very well be that a viscosity could be assumed which would 
explain largely the inclination and not be particularly unfavorable to the 
eccentricity. There were so many partial contradictions and so much 
uncertainty that Darwin attempted to draw no final conclusion from this 
discussion (3, p. 879). 

Another question, still more critical, is the distance from the earth to 
the moon when the day and month were of equal length. In 2, section 18, 
neglecting part of the action of the sun, Darwin found that the day and 
month were equal at 5** SO"", corresponding to a distance between the centers 
of the earth and moon of 10,000 miles. In 3, section 22, including all the 
action of the sun, he found that the initial period and distance were less. 
The numerical results were not obtained, but he stated (3, section 22, p. 836) : 

It 18 probaUe that an accurate solution of our problem wouki differ oonsiderably from 
that found in "Precession" (5^ 3$™), and the oonunon angular velocity of the two bodies 
might be very great. 

In the summary of this same paper, p. 877, he said: 

In section 22 it [the sun's action] is only so far considered as to show that when there 
is identity of periods of revohition of tiie moon and earth, the angular velocity must be much 
greater than that given by the sdution in section 18 of "IVecesBioa." 

Computations given at the end of the present paper, section 14, show 
that the difference is actually unimportant. 

Apparently, in order to make this fission hypothesis workable it must 
be shown that if the day and month ever were equal they had such a period 
that the distance from the earth to the moon was much less than 10,000 miles. 

There are many places where it would be easy for a careless reader 
of Darwin's work to lose the connection between the conclusions and the 
hjrpotheses upon which they were based. For example, taking approxi- 
mately that viscosity which would produce the most rapid tidal evolution, 
he found ^ that 57,000,000 years ago the day was 6^ 45" long, and that 
the length of the month was 1.58 of our present days. Notwithstanding 
the fact that this is quite a different thing from having proved that 
the earth-moon system has actually gone through this series of changes, 
undoubtedly many first-hand and more second-hand readers of Darwin's 
work have suppoiBed that this computation gives a fairly certain and 
definite account of the evolution of these bodies. But it is interesting to 
find in the same memoir, section 14, under the hypothesis that the observed 
secular acceleration of the moon's mean motion is due entirely to tidal 
friction, and also that the earth is purdy viscous^ the conclusion that the 
length of the month is now being increased at the rate of only ^ 20" in 

> 2, Section 15, Table IV, and pp. 529-531. 


100,000,000 years. Very different results were obtained by assuming that 
the earth is dastiahviscous. Under this assumption in 700,000|000 years 
the day will be about as long as at present, the month nearly a day shorter 
than at present, and the obliquity of the ecliptic about 6^ less than it is 
now. The discussion of the secular acceleration of the moon was closed 
with these remarks (2, p. 483, last paragraph) : 

The ooncluflion to be drawn from all these calculations is that, at the praent time, 
the bodily tides in the earth, except perhaps the fortni^tly tide, must be ezoeeding^y 
small in amount; that it is utteily uncertain how much of the obs^ved 4' of aocelnation 
of the moon's motion must be referred to the moon itself, and how much to the tidal frictkm, 
and accordingly that it is equally uncertain at what rate the day is at present being 

Notwithstanding these uncertainties, in the general discussion at the 
end of the final paper of the series (4, pp. 632-533) Darwin states: 

The previous papers were prindpaUy directed to the case of the earth and moon, and 
it was there found that the primitive condition of those bodies was as follows: The earth 
was rotating, with a period from two to four hours, about an axis inclined at 11^ or 12^ to 
the normal to the ecliptic, and the moon was revolving, nearly in contact with the earth, 
in a circular orbit coincident with the earth's equator, and with a periodic time only slightly 
exceeding that of the earth's rotation. 

Then it was proved that lunar and solar tidal friction would reduce the system from 
this primitive condition down to the state which now exists by causing a retardation of 
terrestrial rotation, an increase of Ixmar period, an increase of obliquity of ecliptic, an 
increase of eccentricity of lunar orbit 

It was also found that the friction of the tides raised by the earth in the mocm would 
explain the present motion of the moon about her axis, both as regards the identity of the 
axial and orbital revolutions, and as regards the direction of her polar axis. 

Thus the theory that tidal friction has been the ruling power in the evolution of the 
earth and moon completely coordinates the present motions of the two bodies, and leads 
us back to an initial state when the moon first had a separate existence as a satellite. 

This initial configuration of the two bodies is such that we are almost compelled to 
believe that the moon is a portion of the primitive earth detached by rapid rotation or 
other causes. 

The problem of tidal evolution is an extremely complicated one and 
the uncertain factors which enter into it are very many. Darwin's treat- 
ment of it as a mathematical problem was masterly and worthy in every 
respect of the highest admiration. He was generally very cautious in 
drawing conclusions with respect to the actual earth-moon system. The 
danger lies in the 'formidable and protracted analyses coming in between 
the hypotheses and the conclusions, which might lead one to suppose that 
results drawn from a particular set of postulates necessarily belong to the 
earth and moon, particularly if they, in a general way, coincided with his 
preconceptions of cosmogony. Even though Darwin may have been with- 
out fault in this respect, it is not certain that less critical minds, especially 
if they were without the illuminating experience of finding by actual com- 
putation how great changes in the results would be produced by admissible 
changes in the hypotheses, would not attach imdue importance to some 
particular computation. There is nothing deduced from observations so 
far made or from Darwin's investigations that would prevent one, if it 
suited his fancy, from drawing the conclusion that the motions of the earth 
and moon have been for 100,000,000 years about as they are at present. 


In queBtioQS of cosmogony, where immense intervals of time are in- 
volved; the problem of tidal evolution is obviously one of great importance, 
unless it shall some time be shown that it is not a sensibly efficient factor. 
The two most obvious methods of determining its efficiency are by direct 
attacks from the mathematical standpoint, or by comparing its certain im- 
plications with as many facts given by observation as possible. The first 
is mainly the method of Darwin, and he has written what will certainly 
always be an extremely important chapter in the question when considered 
in the broadest possible way. His results can be improved, apparently, 
only by a determination of the physical properties of the earth as a whole, 
and by an estimate of the loss of energy in the ocean tides. While there 
is hope for the former from seismic vibrations, certain astronomical phe- 
nomena, and the character of the crustal deformations as revealed by 
geological studies, the results are not now so well established that they do 
not need support from other sources. The second method, that of com- 
paring the positive implications of the tidal theory with observed facts in 
as extended a way as possible, is broadly speaking that adopted in this 
paper. Since there can be no test of time-results except on the basis of 
other doubtful hypotheses, and since it is impossible to draw any certain 
conclusions in the questions involving the time, this variable has been 
entirely eliminated from the discussion except in section 14. In a general 
way it may be said that the energy of the system has been taken as the 
independent variable, for it is known that under any sort of friction it 
must degenerate into heat. The results are characterised by certainty 
so far as they go, but as compared with Darwin's they are in most oases 
much less explicit as to particulars. The discussion is mostly attached 
to the fundamental equations of moment of momentum and energy. After 
the work was well advanced it was found that Darwin had applied fun- 
damentally the same methods to illustrate his results in a paper supple- 
mentary to his main series and published in a diiBTerent serial (No. 5 in the 
list previously given). The variables he used were different from those 
employed here, but, though for certain purposes they may be more conve- 
nient, nevertheless, for the sake of complete independence, those originally 
selected have been retained. 

When the discussion is based simply upon the moment of momentum 
and energy equations, the number of quantities to be determined is greater 
than the number of the determining equations. To attain the greatest 
simplicity the general problem has been divided up into a number of 
special cases covering altogether the entire field. In this way each special 
problem is very easily understood and the question as a whole is much 




The problem treated will be that of the tidal interactions of m^ and m^, 
which, in certain cases, are supposed to be disturbed by a large distant 
mass is. The masses m^ and m, will be assumed to be spherical. Then let 

a^^ radius of m^; a, » radius of m, 
r«- distance from the center of m^ to the center of m, 

/>.«« — r» distance from m^ to center of gravity of m, and m, 

/>,— ^ — fB distance from m, to center of gravity of m^ and m, 

X, y, g^B&i of fixed axes with origin at center of gravity of m^ and m, 
i ""inclination of orbits of m^ and m, to xy-plane 
ii"- inclination of plane of equator of nii to xy-plane 
{,»i inclination of plane of equator of m, to xy-plane 
e^i eccentricity of relative orbit of m^ and m, 
a » major semi-axis of relative orbit of m^ and m, 
B angular velocity of revolution of m^ and m, 

P» ^ -B period of revolution of m^ and m, 

a>] B angular velocity of rotation of m^ 
a>s-" angular velocity of rotation of m, 

Di— -= — -"period of rotation of m. 
* a>i 

!),—-= — —period of rotation of m, 

iS "- a distant disturbing mass 

r^«" distance of S from center of gravity of nti and m, 
jP» period of revolution of S 
M, M\ M^— whole moment of momentum of system, exclusive of S, 

about the z, x, and y-axes respectively 
£■> whole energy, both kinetic and potential, of system exclu- 
sive of S 

The moment of momentum of m^ about the z-axis is the moment of 

momentum about its axis of rotation multiplied by cos t\ plus its moment 

of momentum of revolution around the z-axis. The rotational moment 

of momentum of tn^ around an axis through its center and parallel to the 

f-axis is . 

Cifnfi^Wi cos t| 

where C| is a constant depending upon the law of density. The corre- 
sponding quantity for m, is 

c^mTfl^a)^ cos t. 

The moment of momentum of revolution of m^ about the 2-axis is 

m, pi' cos i — 7 — \~^r,r^ cos i 



and the oorresponding quantity for m, is 

m, pJ cos i — 7 — \ ^ .^ T^O cos % 
Therefore the total moment of momentum about the 2-axis is 
ilf = — ^^-r^O cos i-^cMjiit^a). cos it+e.mMJo). cos u 

The whole energy of the system is the kinetic energies of rotation, plus 
the kinetic energies of revolution, plus the potential energy. 

The kinetic energy of rotation of m^ is ic^mia^^oji^, and there is a similar 
expression for the kinetic energy of rotation of m,. 

The kinetic energy of revolution of m^ is 

and there is a similar expression for the kinetic energy of revolution m^ 

The potential energy of the system is ^-^, where k^ is the gravita- 
tion constant. ^ 
Therefore the total energy is 

From the two-body problem we have 

Hd=±*^(m,+m,)a(l-e») (3) 

the determination of the ambiguous sign depending upon the direction 
of revolution, and 



By means of (3), (4), and (5) equations (1) and (2) reduce to 

„_ +m,OT,fc«P* r; — :; .• . 2m:^m^a^* cos t, . 27rc,wt /i,' cos t, . 

^-(2;r)»(t»,+»g»V*-' «'»*+ D, + D, <®^ 

the sign of J^ depending upon the direction of revolution, and the signs of 
D| and D, upon the directions of rotation of m^ and m, respectively. 



The units are so far arbitrary. We shall choose them so that 

unit of time = mean solar day 

Then equations (6) and (7) become 

X Di Cj \o,/ Vf 




If we represent the ascending nodes of the phuies of the orbit and of the 
equators of m| and m, by Q, Sii, and Q, respectively, the moment of 
momentum equations for the x and y-azes are similarly 

M--p».Ji:::^sin*sina+ '"'°^!;°^^' +g?fty "^°^^°^^' 

>^« ni n 7 • • /-> , TO. sin t, cos Q, , c,/a,\* m, sin*. COB ft, 

M'-P»^l-c»smtcoBft+-* ^^ »+-|(^) -=i ^ 



In treating the rotations of such bodies as the earth it is not permissible 
to regard them as homogeneous, for in the case of the earth the density 
of the snrface rock averages about 2.75, while the average density of the 
whole earth is 5.63. If we let a represent the density of the sphere m, 
the well-known expression for it suggested by Laplace is 

— ; — . (11) 



where r is the distance from the center of m, and where and /i are con- 
stants depending upon the constitution of the body. According to this 
law the density of the body increases from the surface to the center, and is 
finite at both the surface and the center. We shall determine O and /i by 
making both the surface density and the mean density agree with the 
results furnished by observation. 

The mass is found from the equation 

— 4jr / m^dr — 4to"G / — sin (;*— ) d — — — 3~Csin /i— /i cos /i] (12) 

Let a^^ represent the surface density and a the mean density. Then 
and fi are determined by the equations 

<7»^ -<? sin /I a;£> - 3G [sin ju— /i cos /i] (13) 

The density at the center is 

a'^'^Qfi (14) 

In the case of the earth a^^ « 2.76, a^ — 5.63, whence it is f oimd from (13) 
and (14) that 

<?- 4.39633 /i- 2.46679 (7/^^-10.840 (16) 

In our ignorance as to the density of the surface material of the moon, 
m^ we shall assume that for it /i,—/! and determine (7, from (12) so that 
81.7m, -m,. We find, taking Oj- 3,958.2 miles and a, -> 1,081.5 miles, that 


a, -3.32 









Letting / represent the moment of inertia of the sphere m, we have 

/= / r*cos*^dm= j j j ar^coa^fpdrdfpdO (18) 


Substituting the expression for <r from (11) and integrating, we find, 

/-^^[3 (/i»-2) sin /I-/I (/i»-6) cos /i] (19) 

or, making use*of (12), 

J ^ 2aHn [3 (/i^— 2) sin /i-/i (/i»~6) cos p] .^ 

3p? sin fi—fi cos fi 

Hence the values of c^ and c,, occurring in (1), are determined by equations 
of the form 

^ ^ 2 [3 (/i»~2) sin pt^fi (/i»-6) cos ft] .^n 

3/1* sin fi—pt cos pt 

When m^i represents the earth and nt, the moon, we find from this equation 
and the value of /i given in (16) that 

c,-c- 0.33594 (22) 

instead of 0.4, the value for homogeneous spheres^ 



V. SPECIAL CASE i«0 i\"0 a,-0 e»0 ^^O.'^ 

This means that two bodies undisturbed by any exterior force revolve 
in circles, that the radius, mass, and angular velocity of rotation of one of 
them are so small that its rotational momentum and energy may be 
neglected, and that the axis of rotation of the other is perpendicular to the 
plane of their orbit. In this case equations (9) and (10) become, writing D 
in place of D^, 

ilf=pi+:^ (23) 

^ D 


We may choose the direction of revolution of the bodies as the positive 
direction. Then only a positive P can have a meaning in the problem, 
since a revolution in one direction can not be reversed without a collision 
of the bodies. D is positive or negative according as the rotation is in the 
same direction as the revolution or the opposite. Under the hypotheses 
adopted M is rigorously constant. When Z) =■ + oo then P « AP; when 

D-^then P«0; when D=Ztm(0+0 then P«-oo; when D«Zim(0— t) 

then P = + 00 ; when D — — oo then P = HP. Consequently the curve 
defined by (23) is as given in fig. 9. The part of the figure to the right of 
the P-axis belongs to the case where the rotation of m^ and revolution of 
m, are in the same direction, and the part to the left where they are in 
opposite directions. 


Fio. 9. 

The slope of the curve, or the ratio of the rate of change of the period 
of revolution to that of rotation, is found from (23) to be 

dP / dP 
dt / dt 


* For a nmilar treatment of this problem see No. 5 of Darwin's papers. 



The values of P and D must also satisfy equation (24). Starting from 
any epoch, P and D must change, if at all, so that E shall decrease, for if 
there is (tidal) friction in the system, some of its energy will degenerate 
into heat and be dissipated. 

The relations between P and E and D and E are found from (23) and 
(24) to be respectively 

AP (26) 





D^ {MD-m^y It 


In these equations P may vary only from to + oo while D may vary 
from— ooto + oo. 

The curve whose equation is (26) has two forms according as E^ con- 
sidered as a function of P, has a finite maximum and minimum or not. 
In case there are a maximum and a minimum it has the form /, fig. 10. 
Since E can only decrease, it follows that if at < «fo ^^ period is on the part 

Fio. 10. 

of the curve ob it will decrease to the abscissa of the point h; if it is at h 
it will permanently remain at that value; if it is between 6 and c it will 
increase toward the value at h; if it is at c it will remain there unless the 
system suffers some exterior disturbance, when it will increase toward h 
or decrease toward d according to the nature of the disturbance; if it is 
between c and d it will continually decrease toward zero. 

When the curve has no finite maximum and minimum it has the form 
//, and then whatever may be the value of P at <«^o ^t will continually 
decrease toward zero with decreasing E, 

It is unnecessary to draw the curve whose equation is (27), for the 
relation between the change in P and the change in D is given in (25). 



It is seen from this equation that when the rotation and revolution are 
both in the same direction P and D either both increase or both decrease, 
if they change at all. 

The necessary condition for a maximum or a minimum of B is, from (26), 

^Plg^Pl-AfP+m.-0 (28) 

The corresponding condition from (27) is 

The only D having a physical meaning is real. Since no real negative D 

satisfies this equation, it follows that when D is negative E has no finite 

maximum or minimum. In this case by fig. 10, dotted curve, the period 

of revolution must always decrease and the two bodies ultimately fall 

together. Taking the last four terms to the right and extracting the cube 

root, we have, 


Since the roots of this equation are the same as those of (28), it follows that 

whtnEis a maximum or minimum D '=> P and the system moves (U a rigid body. 

The real roots of (28) are the abscissas of the intersections of the curves 


y="ilfP— nij 


It is evident from fig. 11 that there are two, or no, intersections of 
these curves. 

. 11. 

For a given ntj the value of M may always be taken so great that there 
wiU be two real roots; or, so small that there will be no real roots. The 
limiting value of M, as it decreases, for which the real roots exist is that 
value for which they are equal. The condition that (28) shall have equal 
roots is 




Since this value of P must also satisfy (28) we have 



M^^m,k (31) 

as the value of M for which there are two equal roots of (28). For greater 
values of M (28) has two distinct real roots. The roots are the abscissas 
of the points 6 and c, curve I, fig. 10. The smaller root corresponds to a 
maximum of £, and the larger to a minimum. When the roots are equal 
the curve has a point of inflection with tangent parallel to the P-axis. 

We may express the rates of change of P and D in terms of the rate of 
change of E by differentiating (23) and (24) and solving. We find 

dP 3P»Z) dE dD PD" dE /jg) 

di "" 2;r(P-Z)) di dt ^ 2m,7t{P-D) dt 

Consider the case first where the direction of rotation and revolution 

of nil are the same, i.e., when D > 0. Since --g- can be different from sero 

only when P%D, then when P>D both P and D must increase whatever 
may be the character of the tides as determined by the physical condition 
of mi,* and when P<D both P and D must decrease. When m^ rotates 
in the negative direction, i.e., when D<0, P must always decrease and 
D always numerically increase. When P'^D equation (23) becomes 


for which, by (28), £ is a maximum or minimum. 

We are supposing the orbit a circle and the axis of tn^ perpendicular 

to this orbit. Consequently when P'^D there can be no change in the 

motion of the system due to the tides. Therefore the right members of 

(32) must carry (P—Dy as a factor, where 7 > 1. The exponent ; can not 

be fractional for then the rate of change of P would be imaginary for P<D. 

Consequently / is 2 or some greater positive integer. The velocity of the 

p D 

tide with respect to the surface of m^ is , and the tidal force is pro- 

1 "^ 

portional to -,. If we assume that the friction is proportional to the height 

of the tide and the first power of its velocity over the surface of m^, and 
that the loss of energy, i.e., the work done against friction, is proportional 
to the square of the friction, equations (32) become 


dP e^P-D 

dD c, D(P-D) 

dt~r* PW 

dt m^r* P 

where c^ is a constant depending upon the physical condition of m^. 


Equations (32) are very instructive, for they prove rigorously, under 
the hypotheses we have adopted in this section, that the rates of change 
of both the day and the month are proportional to the rate of the loss of 
energy, however it may be lost. That is, if tidal friction is now almost 
exclusively in the ocean tides, as Darwin supposed,^ in so far as the earth- 
moon system satisfies the hypotheses of this section it is quite immaterial 
whether the energy is lost in the manner described by the spherical har- 
monic analysis when applied to the viscous theory, or whether it degen- 
erates after the waves have been time after time reflected from the con- 
tinents and have run into narrow bays or into the high latitudes. The 
relation of the tidal wave to the moon is not directly involved as it is in the 
elementary geometrical discussions of tidal friction, though of course the 
rate, and therefore the phase, of the friction depends upon the viscosity 
of the water. This would increase one's faith in the spherical harmonic 
analysis for such an earth and ocean as we have if it were not for the fact 
that the irregularities in the depth of the ocean and in the outlines of the 
continents undoubtedly greatly change the whole amount of friction. 

The moon sets up motions in the waters of the ocean, but not all of the 
energy possessed by this water is lost. At the succeeding disturbance of 
the same region by the moon the phase of the tidal deformation still per- 
sisting may be such that the moon's attraction will tend to destroy it rather 
than generate a new wave; or, the phase may be such that the moon will 
augment the tide. On a world covered with oceans of many dimensions and 
depths we should expect to find places where the natural periods of oscilla- 
tions in water basins are such that the moon's disturbance builds up consid- 
erable tides, and others where they are kept low. In the former case the 
friction of the water prevents their becoming excessively large; if the water 
were entirely frictionless they would increase until the resulting alteration in 
their period would lead to their destruction by the moon's disturbing forces.' 

One method of finding the present rate of tidal friction, at least so 
far as it is due to ocean tides, is to compute from tidal observations in all 
parts of the earth, and from the frictional properties of water, the actual 
waste of energy.' If one were to observe the energy manifested when the 
tide rims through a strait on our coasts, he would be apt to overestimate 
the work the moon is doing upon the earth. In the first place such condi- 
tions are quite exceptional, and in the second place only a very small part 
of that energy degenerates into heat. When the run of the tide ceases the 
kinetic energy has very largely become potential, and it becomes kinetic 
again when the tide runs out. If the outgoing tide has the same energy 
as the inflowing tide there has, of course, been no loss, and, according to 
equations (32) and (33), there is no tidal evolution in such a system as 
we are considering in this section. 

In the units employed the kinetic energy of rotation of m^, the kinetic 
energy of revolution of mj and m, about their common center of mass, 

' 2. pp. 483-484. 

' For a discussion of the observational evidence see Harris, U. S. Coast and Qeod. 
Survey (1900), app. 7, pp. 53^600. Also Chamberlin's paper, ante, pp. 5-60. 
' See paper by Macwillan, ante, pp. 71-75. 


and the potential energy of ntj and m, are respectively 

(rotation) ^^^ (revolution) ^ (potential) _23r ,,-. 

Then from equations (32), (33), and (34) we find 

(rot.) (rcT.) (pot) 

dEmi P dE dEm^-k-mt D dE dEm^+mt -"2P dE r^g) 

di "P-D di di ^ P-D dt di " P-D di 


(rot) / (rev.) (rot) / (pot) 
dEm^ / dE(Mi-\-mt) ^ , P dEmi / dEmi^mt JP (36) 

di / dt ^D dt / dt '^ 2D 

When the directions of the revolution and of the rotation are the same, 
the loss of energy of rotation is to that of revolution of both bodies as 
the period of revolution is to that of rotation, and the potential energy 
gains twice the loss of the revolutional energy. 

The number of periods of rotation in one of revolution is, from (23), 

When (28) is satisfied iV -> 1. The maximum value of iV is defined by 

ni,^-(M-|pi)-0 (38) 

whence, at maximum N, 

'-g^ "-m <»> 



This discussion neglects the rotational momentum and energy of the 
moon, the eccentricity of the moon's orbit, the inclination of the equator 
of the earth to the moon's orbit, the slight oblateness of the earth, and the 
disturbing action of the sun. These factors have probably been of slight 
importance in the series of changes which P and D may have undergone. 
The observations show that at present, taking for the radius of the earth 
that which would give it a volume equal to that of the actual oblate earth, 

a^ » 3,968.2 miles 8 « 332,000 m^ 

a,» 1,081.6 miles D =0.997270 mean solar day 

r' - 92,897,000 miles P « 27.32166 mean solar days 

m| «81.7 m, P' « 366.26636 mean solar days 


From the two-body problem we have ^ 

P'^S+m^+m^ ^*^^ 

Consequently the second equation of (8) becomes 

With the data of (40) the first equations of (8) and (42) give 

imit of time "mean solar day a^ » 0.688303 1 

unit of length « 1 .46286 a^ m^ - 0.62631 1 (43) 

unit of mass « 1.69666 m^ 

Using (40) and (43), equations (23) and (24) become 

M - 3.01187+0.62803-3.639901 .^. 

E - -0.34632 + 1.96689 - 1.61067 J ^ ^ 

It is found from (26) that at the present time the ratio of the rate of 
the change of the month to that of the day is 

f/f-.7.U (45, 

With the value of m, given above we find 


^ There m, of coune, a correspondmg equation for the motion of the moon about the 
earth, but nnoe the direct perturbing action of the Bun increaaes the period for a given 
distance of the moon, the k found from this equation would be too smalL 


Since M is greater than this number it foDows that (28) has two real roots. 
It is easily found by approximation processes that the roots of (28) are ^ 

Pj -0.20535 days =4.9284 hours P, -47.705 days (46) 

Since the present value of P lies between these limits it must alwajrs remain 
between them and continually approach P,. From the formula 



w -v Wl| + t?l J 

where k is the Gaussian constant, and the units are the mean solar day, 
r', and S, we find that the distances fi| and R^ corresponding to P^ and P, 

*~ fii =9,194.35 mUes fi, =345,355 miles (47) 

The maximum possible number of days in a month is at once found 

£«»m(39)tobe iV-29.659 (48) 

The corresponding length of the month expressed in terms of present 
mean solar days is, from the first equation of (39), 

P= 20.345 (49) 

Since the month is increasing and now greater than 20.3 days, the system 
is already beyond the condition of maximum number of days in a month. 
Let us apply these results to Darwin's hypothesis of the separation of 
the moon from the earth by fission, remembering, of course, that a number 
of factors involved in the actual case have been omitted. At the time of 
separation their periods of rotation and of revolution about their center of 
gravity must have been equal. But the solution shows that they moved as 
a rigid body when the surface of the moon was 9,194.4— (3,958.2 + 1,081.5) 
-> 4,154.7 miles from the surface of the earth, which contradicts the hypoth- 
esis that they had just separated by fission. But this is neglecting the 
earth's oblateness, which must have been great. To get an idea of the 
possibilities let us examine a number of modif3ring hypotheses. First let 
us suppose that the earth was then so oblate that its equator reached to 
the moon, and that the law of density was such as to keep its moment of 
momentum and volume unchanged. Then we find that 

equatorial radius = 8, 112.9 miles polar radius = 942.2 miles 

Obviously the spheroid would have broken up long before it attained this 
degree of oblateness, and under the hypothesis that the moment of momen- 
tum was as it would have been in a sphere the equatorial zone would have 
been so rare that one could not account for the matter in the moon. 

In order to avoid the difficulty of the rare periphery, forced by the 
condition on the moment of momentum, we may waive this condition. 
Assuming simply that the earth was oblate, let us find the qualitative 
effects on the initial distance of the moon. The moment of momentum for 

* Darwin in 2, p. 508, taking the earth as a homogeneous spheroid and other data 
somewhat difiPerent, found P, =^ 5.6 h., P, = 65.5 d. 


a given volume will have been greater than that which we have used, and 
we may adjust our formulas for it by increasing the c^ which occurs in the 
original equations (6) and (7). It foUows from (8) that the unit of length is 
now greater than before. Then by (42) the numerical value of m^ is greater 
than with the original e^, from which it follows by (29) and fig. 1 1 that the 
distance of the moon from the earth when the month and day were equal 
was greater than that computed above. This would necessitate an increase 
in the oblateness in order that the earth's equator should have extended 
out to the moon, and the difficulty of having an earth already improbably 
oblate is increased. 

Another hypothesis is that the earth was initially larger and, since the 
separation of the moon, has shrunk to its present dimension. This is 
quite in accord with the general ideas prevailing in the fission theory. We 
ean not apply directly the formulas which have been written down because 
a change in volume would change the distance at which the system moved 
as a rigid body. Consider an instantaneous change of any extent in the 
radius of the earth. This does not change its rotational moment of mo- 

mentum. Then if we employ the same units -=^ is not changed; that is, D 

is changed so that when the new coefficient of m^ as defined in (8) and (42) 
is used the quotient is constant. But an increase in the size of the earth 
would result in an increase in D. Therefore m^ is increased to km^ and the 
condition for equality of the day and month is 

From this equation we find 

dP m^ 
dk "M-4P* 

which is positive for the smaller root of P^. Hence, if the earth has shrunk 
from larger dimensions, the earth and moon moved as a rigid system at 
a greater initial distance than that found above. That is, the hypothesis 
that the earth has shrunk only adds to the embarrassment because of the 
initial great distance of the moon. 

We may try the hypothesis that the moon separated from the earth at 
a distance of 9,194.4 miles, that the earth's law of density was such that 
at that time its radius was equal to this number, and that the moment 
of momentum of the earth's rotation was the same as if its density were 
as it is at present. The latter condition is necessary, for the whole moment 
of momentum is unchanged by contraction. This hypothesis amounts to 
simply attempting to change the law of density as well as the volume so 
that the implications of the hypothesis shall be reasonably satisfied. 

We shall suppose the density is expressible by the Laplacian law, only 
with different values of G and /i from those which are used above. Letting 
/{|-* 9, 194.4, the moment of momentum of the whole system was 

M-2«c'(«.+m,)^!-?^'(l+2)(|4?l-lVa,. (60) 



where </ depends upon the new law of density. We have in the units defined 
in (8) 

P,Af /3,958.2 V nnon,., /e,^ 

the Ci, -^, m^, M, P^, and R^ being given in equations (22), (40), (43), 

(44), (46), and (47) respectively. 

With this value of c' we must determine a new value of /c, say fi', from 
(21). To facilitate the solution (21) may be written in the form 



3 ' ju'» ' 3 (tan /— /) 
We may draw the graph of this function. It is the sum of two functions 


J'' "3 -''-/' 

4 tan ft' 



Fio. 12. 

For /i'—O we see that yi= — ooand y,« +oo, but that Vi+y,— 2— c'«1.98. 
The curve for y^ is very simple. The value of y^ is positive while j/ varies 
from to ;r, vanishing at /i^«;r. Then y, becomes negative and remains 
negative until y,— — oo by the vanishing of tan /£'— /i' at j/^257^ 30'. 

4 Stt 

Then y, changes sign and descends from 4- <» to - at ^m' « — and to zero 

3 2 


for fi'^2n. Then it decreases to — oo in the fifth quadrant, changes sign 
to +00 and decreases to zero at p!^Zn and to — oo again in the seventh 
quadrant. In a general way this cycle of changes is repeated indefinitely, 
the points where y^ becomes infinite approaching nearer and nearer to 

)u^»(2n + l) ^, n being an integer. The curves y^, y, are given in fig. 12. 

Both yx and y, being even functions of pi the curves are entirely to the 
right of the vertical axis. 

From the diagram it is seen that the only places where Vi+t/, may 
vanish are at the left of 9r as at a, to the right of ;r, as at b, to the right of 27r, 
as at c, and in general to the right of n;r, n any integer. There are, in short, 
an infinite number of determinations of p!. But when we consider that 
the law of density is 



we see that if n<ik' <2n the density will be negative near the surface and 
elsewhere positive; if 2n<p!<Zn there will be a single spherical layer 
between the center and surface where the density will be negative; and if 
nn<ii'< {n + \)n there will be n layers of negative density if n is odd, one 
of them being at the surface, and n— 1 layers of negative density if n is even, 
the surface density being positive. Consequently in considering such a prac- 
tical question as the separation of the moon from the earth, in which nega- 
tive densities would have no meaning, we need consider only the possibility 
of a solution to the left of n. The value of fi', for which y^ vanishes, is 


From fig. 12 it is seen that Vi+y, can not vanish between this point and n. 
It is easily verified that no value of /i'< 142.5^ will satisfy (52). For 
example, we find the following corresponding sets of values 


/i'- 0^ 46^ 90^ 135° 140° 
/(/!') -0.58 0.40 0.38 0.33 0.32 

Consequently the smallest fi' satisfying the conditions is greater than n, and 
the hypothesis of the separation of the moon from the earth requires, so far 
as the factors and the law of density here considered are concerned, that we 
assume that the surface density of the united mass was negative just previous 
to the separation. If we had used the oblateness of the figure of the mass, 
a still larger /c' would have been found. However, it is not impossible that 
neglected factors may somewhat relieve the theory of these embarrassments. 
Thus we see that when we add any of the hypotheses of an original 
oblateness, shrinking, or different law of density singly, the difficulties of 
the hypothesis are not relieved. 


VII. CASE t=0 t\-0 t,-0 a,+0 e-0 5-0.* 

This case differs from V only in that the rotational momentum and 
energy of m, are not supposed to be zero. In this case equations (9) and 
(10) become, supposing that c^^Ci, 

In these equations only P, D^, D^, and E can vary, the last decreasing 
through loss of energy by friction. If by means of (55) we eliminate one 
of the variables from (56) we have a relation among the other three. This 
equation may be considered as defining a surface. Let the £-axis be pointed 
upward. Then starting from any point on the surface the variables other 
than E may change in any way, so far as these considerations show, so 
that the point descends. 

We shall now find the maximum and minimum values of E. In order 
to simplify the algebra let us put 

Using this substitution and eliminating P between (55) and (56) we obtain 

g. -1 xJ^+J!L (58) 

t: [M—u—vf m, icm| 

The necessary conditions for a maximum or minimum of E are 

mJM— w-vr dE 

27t du 

=— m^+u [M—U'—vf^O 

— ^2^ --^=-tcmi+v[M-'U-vf^O 


Multiplying the first equation by v, the second by u, and taking their 
difference, we have ku=^v; whence, by (57), 

D, = D, (60) 

Then the first equation of (59) becomes 

-— *+[M-(l+ic)u]»=0 (61) 

♦ See 5, pp. 178-181. 

CA8B vn. 101 

which, after extracting the cube root, is 

Af-(^')* + (l+/c)tt 

Let the common value of D^ and D^ be D. Then this equation becomes 

M -Z)» -f (1 -f ic)^» -D* +[m,+(^^Jm,}^ (62) 

But when D| — Z), «* D, (55) becomes 

M=P»+K + (^|)m,>i (63) 

where P^ must be taken with the positive sign. Since (62) and (63) must 
simultaneously be satisfied for a maximum or minimum, we must have, 
when £ is a maximum or minimum, 

P-D,-Z), (64) 

That is, the energy is at a nutximum or minimum only if P ^ -h D^'^ -h D^, 
i.e., if the whole system moves m a rigid body. 

There can be a maximum or minimum only if (62) has real roots. The 
treatment of this question is the same as that given in IV, except that we 

must replace m^ of that section by m^ + f — ] m,. Hence the condition for 
real roots, and therefore for a maximum and a minimum of E, is, by (31), 

M^^[m, + (^|)' m,l* - A(i ^^)krn,^ (66) 

Let us suppose the inequality (65) is satisfied and then consider the 
surface defined by (58). It will be most convenient to give E a series of 
constant values and to draw the corresponding equi-energy curves. To 
simplify the treatment let 

w^u-^v (66) 

Eliminating u from (58) by means of this equation and solving for v we find 

'Lti,.y,±jz^:;^^K^M^K (67) 

K M K K{M--Wy KK 

Consider first the function 

^(^)._!^+a+jOg^+ (i+*)'».g (68) 

K K{M—Wy KK 


''-±lv^w±^Jf(^ (69) 


The derivative of (68) is 


2io 2(i[+1)bi , 

" K "^KiM—XD)* 


It depends only upon u> and is negative for all w > M. For w~M it ia 
infinite. For u> leas than, but near to, Af it is positive; when M is large it 
vanishes and becomes negative and again vanishes and beoomea positive, 
as to decreases from JIf to 0. It is positive for all u)<0. From these facts 
and equation (08) fig. 13 is drawn, the curves £„ . . .,Bf belonging to four 
values of E such that ^,>B, >£,>£,. 

From equation (69) and fig. 13 the equi-energj curves in fig. 14 are 
drawn, the correeponding curves being similarly lettered. There are two 
points of particular interest, A and B. As the energy decreases, the energy 
curves descend to a point at A. That is, on a section through A approxi- 
mately parallel to the v£-plane the point A is a minimum. With decreas- 
ing E the energy curves separate at A and recede in opposite directions 
nearly parallel to the to-azis. Hence a section by a plane through A approxi- 
mately parallel to the to£-plane has the point A &s a maximum. Therefore 
A is a minimaz point of the surface. 

The minimax point A, fig. 14, corresponds to the point A, fig. 13, at 
which /(io)-0. Therefore, by (69), for this point 


CA8B vn. 


At the point A, fig. 13, we have also 

ic df(w) 
2 dw 

,(ic+l)mj ^ 

which becomes by virtue of (71) and (66) 


agreeing with (61), the condition for a maximum or minimum (or minimax). 

Fio. 14. 

The point B of fig. 14 corresponds to the point B of fig. 13. At the 
point B| fig. 13, we have 


which lead again to equation (61). Therefore the smaller real root of (61) 
belongs to a minimum and the larger to a minimax. 

However the system may change under loss of energy the w and v of 
fig. 14 must always go from a curve of higher to one of lower energy. When 
the energy is greater than that for which the point A appears in fig. 14, 
the curves give us no positive knowledge regarding the series of changes 
the system may undergo, except that if at any time t0>-ti+i'<Af it has 
always been, and will always be, less than M, and the opposite. But if 
the energy is less than that for which A appears in fig. 14, then, if at any 


time the w belonging to the system is less than the abscissa of A, that is, 
if the point is on the curve E^, it will always remain less than the abscissa 
of A and will approach the point B as a limit; but if the w is greater than 
the abscissa of A and less than Af it will always remain between these two 
values. The abscissa of every point on the oval part of £, is in all possible 
cases less than the abscissa of A, for the points on this branch of the curve 
correspond to the points of the curve E^ of fig. 13 which are above the 
t0-axis and to the left of the line w=*M. This part of the curve is always 
to the left of A, for the curves of fig. 13 are all derivable from any one of 
them simply by vertical displacements. 

A simple case is that in which the two bodies are precisely alike in every 
respect and have at any time similar motions. Then from the symmetry 
of the problem the motions of the two bodies will always be the same. 
Equations (55) and (56) become in this case 

The treatment of these equations is precisely like that of (23) and (24) 
of section V if we replace m^ of that section by 2fn^. 




With the data given in (40) we find 

(?)'^ -icmi -0.000914 m^ (73) 

Consequently the moment of momentum and energy of the earth-moon 
system are respectively 

Af - 3.01 187 + 0.62803 + 0.00002 - 3.63992 1 

f (74) 

E - -0.34632 + 1.96689+0.00001 - 1.61058 J 

The limiting value of M for which a maximum or minimum may occur is, 
by (65), 1.5614. Since the actual M is greater than this quantity there 
are a minimax and a minimum. 

We shall now solve (61) and find the value of u corresponding to the 
minimax. With the values of k and M given in (73) and (74) we find by 
methods of approximation that the value of u satisfying (61) and corre- 
sponding to the minimax is 

u -3.04719 ii;-2u -6.09438 (75) 

The corresponding common period of the system and distance of the moon 
are respectively 

Pi -0.20554 day -4.9329 hours R^ -9,200 mUes (76) 

or more than 42 miles greater than when the moon's rotational momentum 
and energy were neglected. The whole energy of the system for this value 

^'^"^^ ^-10.666 (77) 

At the present time in the earth-moon system we find from the data 
of (40) and (73) and the condition Z),-P that 

u -0.6280257 v -0.0000005 ti;-u+v -0.6280262 (78) 

Since the present energy, given in the second equation of (74), is less than 
that corresponding to il of fig. 13, which is given in (77), and since the 
present value of w given in (78) is less than that corresponding to A, given 
in (75), the energy will continually approach the value corresponding to 
the minimum B, when the earth-moon system will move as a rigid body. 
Although it is thus possible, under the hypotheses, to draw positive con- 
clusions as to the conditions toward which the system is tending, it is, of 
course, not possible to afibm, even aside from all factors neglected in this 
section, that the system ever descended from the condition corresponding 
to A. This discussion simply gives the numerical values belonging to the 
condition corresponding to A, which may or may not have been verified. 
Thus the real question of interest gets no conclusive answer by supposing 
the moon of finite size; the introduction of this factor simply embarrasses 
the fission theory a little more. 



If the earth and moon were derived by the fission of a parent mass, 
the process has presumably been exemplified elsewhere. We shall conclude 
that the earth has had an exceptional origin only as a last resort. It has 
been many times suggested that the binary stars may have originated by 
the breaking up of larger masses, and See especially has urged this view 
and applied Darwin's formulas in an attempt to explain the dimensions 
and eccentricities of their orbits.^ 

We shall apply the methods developed here to the problem. In order 
to simplify it as much as possible we shall suppose first that the parent 
mass divided into two similar and equal masses. We know that this rela- 
tion of the masses of the two members of binary stars is, in a number of 
cases, nearly fulfilled, and nothing is known to make this assumption seem 
improbable. We shall assume that immediately after the fission the system 
moved as a rigid mass. Then the equations for the moment of momentum 
and energy are p i o^ 

differing from those in section V only in that m^ has been replaced by 2m, . 
The condition for a maximum or minimum of ^ is [cf. eq. (28)] 

P*-MP-h2mi=0 (80) 

As has been shown, when M is sufliciently large this equation has two 
real roots. The smallest Af for which there are real roots is that for which 
the real roots are equal. The condition for equal roots of (80) is 

4 i 

M^:^P^=^^6fn, (81) 

Consider the system moving as a rigid body with the two stars in contact . 

Then their orbital moment of momentum will be — times their rotational 
moment of momentum, or ^^ 

P*-^,(andP-i)) (82) 

Consequently the whole moment of momentum of the system is, calling 
this special value of the common period P^^, 

M = (l+cOPo* (83) 

In order that this may be at least as great as the value defined in (81) 

we must have 

c,5i (84) 

The value of c^ depends only upon the law of variation of density of the 
bodies. When they are homogeneous c^^OA. When the density varies 

' Inaugural Dissertation, Berlin, 1892. 


according to the Laplacian law and vanishes at the surface {fflSO^)^ 
we find from (21) that C| « 0.26. The more the matter is condensed toward 
the centers of the bodies, the smaller in general c^ will be. It is apparent 
from these figures that e^ can not be much greater than 0.33, and therefore 
that M can not greatly exceed the value for which there are equal roots. 
Consequently the two roots of (80) can differ but little, which means that 
tidal friction is not competent to drive two equal stars originating in this 
way far from each other. 

Let us suppose Ci >0.33 so that E has both a maximum and a minimum. 
By (82) we have 2m|>-C|Po*. Using the value of M given in (83), equation 
(80) becomes 

P* - (1 +c,)P.^P +c,P.^ -0 (85) 

Since one of the roots of this equation is P«^ we may factor it into P^^P^^ 

P -CjPo*^' -e^P.^Pi - c,Po " (86) 

whose real root gives the minimum value of E. To solve this cubic in 


Then by the theory of cubic equations ' 

Pt -|P.* + ^—i^ -2^(Ci)P.* (88) 


Let us apply these equations to the case of two stars each equal in mass 
and dimensions to the sun. Taking the radius of the sun at 433,000 miles, 
we find from 


that in this case 

Po- 0.2324 

We shall take for e^ the largest possible value, that is 0.4, which belongs 
to a homogeneous sphere. We find from (88) the corresponding largest 
value of P, to be 

P,- 0.307 

which is'the period for a separation of the centers'of the bodies of 1,042,400 
miles. The original separation was 866,000 miles. That is, imder the 
assumption that each of the two components of a binary system is equal 
to the Sim in mass and volume, and that they remain of constant size and 

1 Burnnde and Panton, Thtory of BquaHim$, p. 108. 


shapCi we have proved rigorously that after fission, tidal friction can not 
have increased their initial distance more than 200,000 miles. Such an 
inconsiderable increase as this in the distance between two stars can have 
had no important effects on binary systems. If the system started at the 
configuration corresponding to maximum energy of the curve of fig. 10, 
either it must have gone toward the condition of minimum energy just 
computed, or the two bodies must have fallen together. 

Let us suppose the bodies to have shrunk as a consequence of loss of 
heat so that their period of rotation would have become, except for tidal 
friction, kD, Since the rotational moment of momentum is not changed 
by shrinking, the m^ becomes tcm^, the mass changing its numerical value 
because the definition of units depends upon the dimensions of the bodies. 
Then equations (79) become 

^ ^^ kD n P^^ K'D^ ^^^^ 

Eliminating D we have 


The condition for a maximum or a minimum of £ is 

Pt-MP+2/cmi=0 (91) 

From the derivative -r-^-rrn — h- it follows that, for given values of M 

and nil, the smaller k is the farther apart are the two roots of P. Let the 
common initial value of P and D be P^. Then equation (91) becomes, by 
(82) and (83), 

P*-{1 +c,)P,iP+fx,P,^ =0 (92) 

Since this equation is homogeneous in P and P^, its solution is of the form 

i'*=/(C|,/c)Po* (93) 

For a given amount of shrinkage of a body the constant k is determined, 
because the moment of momentum is not changed by a decrease or increase 
of volume. Hence, if we assume an initial Po and the #c, we may determine 
the final (and greatest) P from (92), and compare the results with the data 
given by observations of double stars. Or more simply, we may assume a 
final P in accord with the data furnished by observations and compute the 
K from 

for various assumed values of Pq. Since the two stars are supposed to have 
been initially in contact, the initial density may be expressed in terms of 
Po, and the final density is determined by the initial density and the amount 



of shrinkmg, which is measured by k. Therefore the final density may 
be determined, by the use of (94), in terms of the initial density and the 
final period. 

We shall assume that P»100 years, or approximately the period of a 
Centauri and f Scorpii. There are many binaries known with much longer 
periods than this, as well as many with shorter periods. Since they have 
presumably all originated in a similar manner, any correct theory must 
explain the long periods as well as the shorter ones. We shall assume that 
the bodies have always remained homogeneous, whence C|»0.4, which 
has been shown to be most favorable to the theory of a large increase of 
period through tidal friction. From the formula for the period in the 
two-body problem we find, using the volume times the density for the 
masses, and assuming that the bodies were originally in contact, that P« 
varies inversely as the square root of a^, or 




The constant of this equation is determined by the fact that the density 
of the sun is 1.41, while in the case of two such stars as the sun we have 
found P^ to be equal to 0.2324. Therefore the constant is 0.2760. 

From the fact that the moment of momentum of a body simply shrink- 
ing must remain constant, we find that 


where a^^ is the initial radius and a^ the final, 
inversely as the cube of the radius we have 

Since the density varies 

-^ — I ^« " W^ -= final density 


By formulas (94), (96), and (97), with P-100 years, Ci-0.4, the fol- 
lowing table has been computed: 









0.2324 day 
36.4 years 

38.8 " 

39.9 " 

43.7 " 

72.8 " 

— 20,660,000 








With an initial density of 1.4 the final density comes out imaginary, indi- 
cating that incompatible conditions have been imposed. That is, it is impos- 
sible for a star having the density and twice the mass of the sun to divide 
and be driven by tides into a binary system having a period of 100 years. 
When cTo— 4X10r^ the final density is infinite, another impossible result. 
When the initial density is 10~^® we find ic«1.03, indicating an expansion 
instead of a contraction. The only initial densities compatible with the 
assumptions lie between 4 X 10~^^ and 10"^^ Cionsequently a double star 
having two equal components with combined mass twice that of our sim and 
a period of 100 years could not originate by fission except when it was in 


the nebulous state with a mean density not exceeding j^ of that of our 

atmosphere at sea-level. On the other hand, if the close binaries revealed 
by the spectroscope have originated by the fission of a single star, their 
period can never become great through the effects of their mutual tides. 
It follows from (92) that this relation between the initial period and final 
period depends only upon the law of density and the amount of the con- 
traction, and is entirely independent of the mass of the system. 

The question may be raised whether the results would not be less 
unfavorable to the theory if the mass of the system were unequally divided 
between the two stars. The discussion of section VII shows that the problem 

remains in a general way the same, 2m| being replaced simply by m^ + f ^j m,. 

We must conclude from this discussion that approximately equal 
binary stars with long periods can have originated by the fission process 
only when the parent mass was yet in the nebulous state. In fact, it 
removes the chief support of the belief that there is any such thing as fis- 
sion among the stars simply because of rapid rotations. From other con- 
siderations Jeans ^ has arrived at the same conclusion. 

>The Afltrophysical Journal, XXII (1905), p. 101. 

CASE X. Ill 

X. CASE t-0 t\«=0 t,=0 a,»0 e+O S-0.* 
In^thifl case equations (9) and (10) become 


m, E 1 . m 

'"V^+^ T- ^+P w 

Eliminating D and P in turn between these equations, we get 

IT pi 

Sohringtorjt we obtain 


Vp.-[«±v¥^ «-V>-D-f)^^^%^ ow 


The'conditions for a mazimum or a minimum of the first of (99) are 



The first equation may be satisfied by the vanishing of any of its three 
factors. P^— is a phsrsical impossibility and makes E infinite. Setting 
the second factor equal to zero and substituting in the second equation 
we have m|— 0, which is impossible. The remaining possibility is e— 0, 
and then equations (101) reduce to 

u-0 Fn-MP+m^^O (102) 

the second equation being precisely the same as (28). 

Similarly the second equation of (99) gives as the conditions for a 
maximum or a minimum 

e -0 --Z)*(AfZ)-mi) + MD»-mi(AfZ)-.mO'-0 (103) 

The latter of these equations gives 

Dt-MD+mi-O (104) 

which, together with the second of (102), shows that for either a maximum 

or a minimum 

c-0 P-D 

* See 3, parts V and VI, and the appendix, pp. 88^-801. 


We shall now study the surface defined by the second equation of (100) 
by considering the curves for various values of E. Cionsider first the curves 
defined by 

y.-l-5t(lf-^y (105) 


^.^ („_».) („_^) <,„„ 

we find the positions of the maxima and minima of yi and we can then 
easily construct its graph. 

We must now consider the curves defined by equation (106) for various 
values of E, both positive and negative. They all pass through the point 

y, » 0, Z> « -jr^, and are tangent to the Z>-axis at this point. All the branches 

of the curves are asymptotic to the lines D"0 and y^^EM^Iitt except in 
the special case when £»0. 

The sum vyi+y^ must then be considered. Whatever the value of 
Ef all of these curves are asymptotic to D«0 and y» — oo . All of the 

curves pass through the point 2)«Wy y°"ly c^d their slope is sero at this 

point. Their slope vanishes at the points defined by 

The solutions of this equation are Z)-» ± oo , D=»r^^, and the two roots of 


the last factor. When E is large the roots are small numerically, one being 

positive and the other negative. With decreasing E the negative root 

recedes to — oo which it attains at E=0. It then becomes positive from 

+ 00 and unites with the other positive root when 


and for smaller values of E the slope is always negative for values of D 
greater than -^^ 

The curve has a point of inflection for this value of E, for which D = 

Those ourves which are t&ngeat to the axis are found by imposing the 
condition (108) and 

y.+y.-^'-O {109) 

The first factors of (108) give no results of interest. Eliminating E between 
the third factor and equation (109) we get 


agreeing with (103), the coaditioos for a maximum or a minimum. If 
equation (103) has no real roots then there is but one intersection of each 
curve with the £>-axis to the right of D~^. When the equation has real 
roots then for certain values of E there are three intersections to the right 
of D'—j-^. As E approaches — oo the curves approach the line ^— W 

between y— and y=- +1. 

We get the final equi-energy curves from 


They are given in fig. 15. The point A is one of the solutions of (103) 
and IB a mlmmax of E considered as a function of D and e. When this 
point belongs to a positive value of E, as it does in the earth-moon system, 
the curves for this and larger values of E are open on the right to infinity, 

because the curves are asymptotic to the lines «=±*/n — 


E decreases until EAf ' — —a the curves close bX +oo, and for smaller values 
of E they are closed ovals until they vanish at a point B on the Z>-axis to the 
right of A. This point corresponds to a true minimum of E considered as 
a function of D and e. On the left of the «-axis the ourves are shaped 


somewhat like parabolas and vanish at ^oo for EM^^^n. They are the 
analytic continuations of the curves on the right, the union being at infinity. 


When K recedes to— QoasJ? = — it reappears at +qo as iS continues 
to decrease. ^ 

It follows from (105) and (106) that y^+y^Kl for all negative values of E, 
Consequently for all points on the closed ovals we have \e\<l, and the 
system can not be wrecked by a collision of the bodies. Therefore, if at 
any time the configuration of the system corresponds to any point on one 
of the closed ovals to the right of A, it will always tend with decreasing E 
through tidal friction toward the configuration corresponding to the point B, 
and the evolution will end with this configuration. Because of the sym- 
metry of fig. 15 with respect to the line €»0, it follows that if at any 
time eaiO it will always remain zero. 

While under certain conditions the system will inevitably progress 
toward a definite configuration, in the general case it is not possible, with* 
out hypotheses as to the physical condition of the bodies, to determine the 
character of the evolution. The question of greatest interest in the present 
connection arises in the case where the conditions lead to doubtful results. 

If the moon separated from the earth by fission and if its orbit were 
originally circular, it would not become elliptic through tidal friction. Since 
the orbit is now considerably eccentric, we must assume that it was some- 
what eccentric at the time of separation. Cionsequently, let us suppose 
the moon has just separated from the earth so that P^D, and, supposing 
that e^O, let us find whether the moon will fall again to the earth or recede 
from it. Since the orbital velocity will have been such that the moon's 
motion will have fulfilled the law of areas, while the rotational velocity 
will have been uniform, there will have been relative motion of the various 
parts of the system, and consequently tidal friction. We are to find the 
effects of this loss of energy on the distance of the moon. 

Under what seem reasonable assumptions we have seen in section V, 
equations (33), that when the orbits are circular the rate of change of the 
month is given by 

dP c P-D 

dt ""r* PW 

We shall now assume that when the orbit is elliptic the rate of loss of 
energy at any instant depends upon the square of the product of the tide- 
raising force and the angular velocity of the tide over the earth. This 
assumption is equivalent to taking the circular case as applying instan- 
taneously to the elliptic case, and omits the lag in tidal conditions due to 
inertia. With this assumption the equation above becomes 

dt r* 01 ^ ^ 

where c' is a positive constant. 

* This result agrees with that found by Darwin in 2, p. 497, eq. (79) after change of 
variables, notation, and proper specialisation of his problem. 

CASE X. 115 

By hjrpothesis the mean value of is equal to at, and we have therefore 
from the two-body problem 


r'^all— e COB cut +^ {1 — 0032(14)+ . . . .] 


Consequently equation (111) becomes 

^--c^e[2coBa;e+ye+^eco82arf+. . . .] (113) 

where e" is a positive constant. The first and third terms produce no 
secular results. The second shows that P secularly decreases. That is, 
under the hypaiheeie thai the lose of energy ie proportional to the equare of the 
product of the vdodty of the tidal vxivt and Ae magnUude of the tide^'aieing 
force, U foUowB that if the moon had separated from the earth and originally 
tiad been moving around it in a dighUy eccentric orbit in a period equal to 
that of the rotation of the earth, then the friction of the tides generated by the 
moon in the earth would have brought the moon back to the earths Since these 
hjrpotheses certainly approximate the truth, we are led to the very probable 
conclusion that the moon can not have separated from the earth in an elliptic 
orbit and have been driven out to its present position by tidal friction. 

Precisely similar reasoning applies to the hypothesis of the fission of a 
star into a pair of equal stars, and is an additional strong argument against 
the soundness of this theory regarding the origin of binary stars, which 
generally have large eccentricities. 

^ Under the amiinpticn that the planet k Tieoous and with diff«rait araroxiiiiatiQDS, 
Darwin's equatiooa led to the fame refult. See 3, p. 864, eq. (202), also 3, p. 878 and p. 891 




Observations show that in the case of the moon's orbit e^ 0.0549. 
Then we find from (98) that in this case 

M- 3.63082 £« 1.61057 

Cionsequently for the moon at present the curves of fig. 15 are yet open at 
infinity, and so far as this discussion goes, the eccentricity may increase 
to unity, and the system be wrecked by a collision of the earth and moon. 
The most interesting question relates to the least possible distance of 
the earth and moon from each other. Suppose at the time of the assumed 
separation the eccentricity of the orbit was very small, as apparently it 
must have been if the bodies separated by fission. Then the configuration 
at the time of separation corresponds to the point A of fig. 15. The abscissa 
of this point is the smaller real root of equation (104), which for the present 
value of M we find to be 

I>«" 0.206008 days -4.944 hours 

From the relation between the period of revolution and the distance we 
find that the distance corresponding to this period is Rq^9,214.0 miles. 
Neglecting the eccentricity of the moon's orbit we found for the initial 
distance 9,194.4 miles. 

But, as was explained in the preceding section, the initial eccentricity 
could not have been sero for the present eccentricity is different from lero. 
The larger it was the shorter the initial period and the smaller the initial 
distance. It will be a liberal assumption to suppose it was «o»0.1, for then 
the initial perigee and apogee distances differed by 1,800 miles. With this 
value and putting D—P we find as the smaller root of the first equation 
of (98) Do — 0.205797d « 4.939 hours. This corresponds to an initial mean 
distance of 9,207.7 miles, not differing materially from that found when the 
initial eccentricity was neglected. 

CASS xn. 


XII. CASE t4:0 t\4:0 a,»0 6-0 5-0. 

If t is to be taken different from lero and subject to change it is neces- 
sary to suppose that t\ may vary also, for all the interactions are mutual. 
In this case equations (9^) become 

Af'-P«BintsinQ + 

nii sin ij sin Q^ 

M'' — F* sm t cos &2 H — * 1^ 


We may choose the xy-plane so that it shall coincide with the invariable 
plane of the system. Then Af' — Af''— 0, and equations (114) give 

Qj- Q+n;r (n— Oorl) 

With this relation equations (114) become 

[pirint±???i^]Bina-0 [P»8int±^jii^]co8a-0 


P*sm»±— ^-g — *— 


In this discussion we shall take D as essentially positive and let % and <i 
vary from to ;r. Then we must take the lower of the two signs in (116), 
or Si^^Si+n. This equation and equations (9) and (10), under the 
hypotheses of this section, give us 



ni • • tn. sm i. 
F* sm * i-yT — ' 

Di • • ^1 cos *i 
•P* cost+ yv — - 

1 ^mj 

pi • Z)» 


Eliminating D and t| from the last of these equations by means of the 
first two, we have 

^ ^+Pi-2McostP» + M« 

n P« 

The conditions for a maximum or minimum of E considered as a func- 
tion of f and P are 


in^bB^ 2mt2 1 2 Jfcosi 
TT dP"3P«'*"3P» 3 Pl "" 


The first equation can be aatifified only by t »0 (the case t — « makes only 
a change in direction of the axes in the final solution). Then the second 
one gives, since P can not equal infinity, as the conditions for a mi^Tirrium 
or a minimum 

i=0 Pt-AfP+nii-O (117) 

Then the first equation of (116) shows that either i|— or I>« oo . In the 
latter case t| is indeterminate. 

To investigate whether the roots of (117) correspond to a maximum 
or minimum we form (at t « 0) 

-^-2MP» irdidp"° Irak's piVs^^^; ^^^^> 

The right member of the first equation is positive for both roots of (117). 
We must consider the function 

For the smaller root of (117) the right member of this equation is negative, 
and therefore B is neither a maximum nor a minimum for this set of values. 
For the larger root of (117) it is positive and the corresponding value of E 
is a true minimum. Whenever the sjrstem arrives so near the condition 
corresponding to this point that the equi-energy curves are dosed they 
remain closed for all smaller values of E down to the minimum, and under 
the influence of tidal friction the system will inevitably approach this 
condition of minimum E, and having attained it will remain there. 

It follows from the second of (116) that for either value of P satisfjring 
(117) we have D^^P. 



There is a difficulty in attempting to determine the position of the 
invariable plane of the earth-moon system, for the plane of the moon's 
orbit is continually changed by the perturbative action of the sun. In 
the course of about 9} years the line of nodes of the moon's orbit makes a 
complete revolution, and the inclination of the plane of the earth's equator 
varies all the way from 23.6^+7^ to 23.6^— 7^ We shall suppose here 
that the sun is not disturbing the orbit of the moon, and that the inclina- 
tion of its orbit to the plane of the ecliptic is sero. Let us take the plane 
of the ecliptic as the original xy-plane. Then the angle between the plane 
of the ecliptic and the invariable plane is given by 

cos to 

j^ P» + J» COS (23.6») 


VM»+M'«+Af- ^p,^2^»^^^(23.6-)+5l 

Since SJj— Si +n we have 

ti-23.6^-t (121) 

In the case of the earth-moon system we find 

»-4*» t,-19.6** (122) 

Then M of (116) becomes Jf- 8.69664. With this value of M the smallest 
root of (117) is P—0.20862. This period corresponds to a distance of 
£—9,364 mfles. However, the initial inclinations could not have been 
* precisely sero, and oonsequentiy the initial distance must have been some- 
what less tlum this amount. But the chief point of interest is that the 
factors neglected in the discussion of section V so far all make the initial 
distance greater than that found in that place. 


XIV. CASE t-0 t'l-O e = o,-0 S+0. 

The sun affects the evolution of the earth-moon system in two ways. 
First, its direct perturbing influence on the moon makes the month longer 
than it would be if the moon were revolving at its present mean distance 
in an undisturbed orbit. In the second place the tides which the sun 
raises in the earth retard its rotation and reduce the moment of momentum 
of the earth-moon system. We shall consider these two influences separately. 

The relation among the sum of the masses of the earth and moon, the 
mean distance from the earth to the moon, and the moon's period is found 
in the theory of the moon's motion to be 

^a«[l+|(|7)' ]=-*Mtn,+m,) (123) 

where P' is the length of the year. With this relation instead of having 

equations (9) and (10) become, when 6»i»ii— a,— 0, 

The disturbing forces which have made these changes in the equations 
are mostly radial, and so far as the radial components are concerned can not 
change the moment of momentum. The tangential components are periodic 
with equal and symmetrical positive and negative values in a period. 
Consequently the M will not have changed under these influences. 

At the time of the supposed separation of the earth and moon P^D, 
and the first of (124) gives for the determination of P 

P*[l-|(py. . . .]-MP+mj-0 (125) 

We find from the first of (124) that the value of ilf is in this case M — 
3.63428. The smaller root of (125) for this value of M is Po"" 0-205760, 
and this period corresponds to an initial distance of R^^ 9,206.2 miles, a 
little greater than that foimd when the sun's action was neglected. 

Consider the direct tidal friction of the sun upon the earth-moon system. 
The sun's tides lengthen D without producing a corresponding change in 
the motion of the moon. Consequently in this case M is not constant. 

Let us assume, as before, that the amount of tidal friction is directly 
proportional to the product of the square of the tide-raising force and the 
square of the velocity of the tidal wave along the earth. It also depends 
upon the physical condition of the earth. Then we have, including the 
tides produced by both the moon and the sun, 

dE rm,» (P-Z))» g»(P^-P)n ,.2ftx 

CASS XIV. 121 

where the first term comes from the tides raised by the moon and the second 
from the tides raised by the sun. Since E can only decrease, C must be 
positive. The factor of proportionality, C, is the same for both since it 
depends only upon the physical condition of the earth. 

Substituting (126) in the second of (32) we have the rate of change 
of the day defined by 

dD C m^ D 

(ft " 2mi;r r* P 


The length of the month is changed by the moon's tides alone, and 
bears a definite relation to the rate of change of the day due to the moon's 
tides. From (32) this relation is found to be 

D" dt "Pt (ft 

When the sun's action on the rotation of the earth is included we have 

From the formulas for the month and year we have 

fry /_mi+n^_y fPy 

Since nii is large compared to m^, and S is large compared to m^, this rela- 
tion becomes with sufGicient approximation 

Then equation (128) becomes 

If we integrate this equation and determine the constant of integration 
by the present values of P and D, and if we then put P — D and solve, we 
shall have as one of the roots the value of P at the time of the supposed 
separation of the earth and moon. But P and D enter this equation in 
such a complicated manner that it is not possible to express its solution 
in finite terms. The critical values of the variables are 

P.0 i,.o P-.>-o :.0(?rg)-@'.„ 

Only the third of these critical values will arise in the applications which 
will be made here. 


In the present condition of the earth-moon system we find that 

It will appear in the computations which follow that as P and D 
decrease in such a way as continually to satisfy (129), this function de- 
creases rapidly until P — 0.25 and D » 0.25 approximately. For 1 ^ P > 0.25 
it is convenient for the purpose of solving (129) to let 

Q-^ + Pi-Af (130) 

With this substitution equation (129) becomes 

di 3 W P^KP-DJ VP7 di 

If we should eUminate D from (131) by means of (130) the result would 
have the form 

^^F(P,Q) (132) 

If we let Pq represent the present value of P, and M the total present 
moment of momentum of the earth-moon system, then at P»Pt we have 
Q»0, and an approximate solution of (132) is 

F (Pfl) dP (133) 

Successive approximations may be found by the series of operations 

This series of approximations approaches the true value of the integral 
provided the upper limit does not pass beyond a point for which F(P, Q) 
has a singularity.^ In the application of these formulas we are explicitly 
limiting ourselves to a region in which F(P,Q) is everywhere regular. 
The integrations can be very easily carried out by mechanical quadratures 
to the desired degree of accuracy. 

Since we wish to trace the system back and find for what values of P 
and D the two variables were equal, we must take P < Po in the integration. 

» Picard, Traiti d'Avudj^, vol. 2, pp. 301-304. 

CA8B XIY. 123 

When the integration has been carried so far back that P— D is small 
enough to make the second term in the right member of (129) the dominant 
one, it is convenient to change from P to D as the independent variable 
of integration. We may now write (129) in the form 

'«'•"■' Z..[(P_D,. + (a)V_D,.(^)*] "'*) 

For P— D=0 this equation does not have a singular point when D is the 
independent variable, for at this point/ (P, D) vanishes. 

Suppose corresponding values of P and D have been found by (132) 
and (130) until P^ and D^ are obtained, and that their difference is small. 
Then P may be expressed in terms of D by an expression of the form 

P~P.-|il<(D-Do)< (136) 

provided the modulus of D-^D^ is sufficiently small. Substituting (135) 
in (134), expanding the right member, and equating coefficients of corre* 
sponding powers of D—D^, we find 





a/(P,, PJ 2m.(P.-D^(4P.-D,) 

^»D.»[(P.-D.)» + g)'(P'-D,)»(J«)] 



d/(Po,D^ 6m.(P,-I),)P.i 

^^* D* [(P.-D.)»+(^)* iP'-D,y (Jj)] 

6m. (P.- D,)' P.I [p,-D.+(^y (P'-D;) (^)*\ 

, _ .TO, 

i5.'[(P.-D.)' + gy(P'-D,)'(fj)Y 





,(P.-D,)^' + 3P,(P,-I),)'*' 

[(P.-i>.)« + 0'(P'-D.)'(Jj)'] ' 


[(P.-i>.V + (^y(P'-D.)«(^)T 

Since these equations are applied only when Pq—Dq is small and D* 
small compared to P^ we have approximately 


m,Do»PoV mj»Do*PoV 

__Po__. (5Po-8Do) . , 
D.CP.-Do)'^^ 3P«(Po-Do)^ 


By successive application of (135) and (137) the corresponding values 
of P and D can be followed until P—D passes through sero. The value 
of P, for which P^^D, can then be determined by interpolation. The corre- 
sponding values of P and D in the following table have been computed 
from equations (130), (131), (135) and (137). The third column gives 
p^ the ratio of the rate of the change in the rotation of the earth due to the 
sun's tides to that due to the moon's tides. 

















































The initial common period of rotation and revolution is found by this 
computation to be 0.20037 day, corresponding to a distance of 9,045 miles. 


This is 149 miles less than that found when the action of the sun's tides 
upon the rotation of the earth was neglected. That is, when the efifects 
of the sun's tides upon the rotation of the earth are included, it is found 
that the possible initial distance of the moon is not materially diminished, 
and that the theory of the fission of the earth and moon is still subject to 
the serious embarrassment of wide separation immediately after the sup- 
posed division into distinct masses. 



As is well known, there is a secular acceleration of the moon's mean 
motion of about M' per century which has not been explained by the ordi- 
nary perturbation theory. It was long ago suggested by Delaunay that 
it may be due to tidal friction, and Darwin has made an investigation of 
the subject in 2, section 14. 

If we accept the tidal explanation, the apparent acceleration of V is 
due to an actual retardation of the moon, the only result possible accord- 
ing to (32), and a greater retardation of the rotation of the earth. Since 
the rotation of the earth is used to measure time, the period of revolution 
of the moon on this basis apparently is accelerated. In making the discus- 
sion we shall neglect the effects of a,, e, 1 1, i, and S. 

Let — AV| be the gain in longitude of the moon in a century, and — Av, 
the corresponding gain in the angular distance of rotation of a meridian 
of the earth. Then we have 

Ar,— AV|-4^ 

dP 2ndB 2n AVj P* AVj 

dt 0* da 0* (lOOPO* 2;r (lOOP')* 

dP _27c^do) 2n Ay, D* Ay, 
(ft " CO* (ft " a>» (100P0*"2;r (lOOP')' 

From these equations and (32) we find 

dD P^D^ V dP 3mjP« 


(ft 27r(Pi-3mi) (lOOP')' * 2;r(Pi-3m,) (lOOP')* 


Representing the value of P at <— (; by P«, integrating the second equa- 
tion, and determining the constant of integration, we have 

P. -p.. +«. (A-^j »g ^^ (,-,.) (140) 

The present rate of tidal evolution of the earth-moon system depends 
upon the forces acting and upon the physical condition of these bodies. 
If we regard the 4^' per century of apparent gain in longitude of the moon 

as due to tidal evolution, we have a measure of the ^ of equations (32). 


We may use equation (140) to compute the time {t—Q corresponding to 
any value of P provided the physical condition of the system, and partic- 
ularly that of the earth, has not changed sensibly in the mean time. The 
data furnished by geology are showing more and more that the earth has 
been sensibly in its present state, except for approximately periodic oscil- 
lations in its climate, for many millions of years. For the purposes of 
computation we shall assume that it has been indefinitely so. While this 
assumption is not strictly true, the actual observational data show that it 
IS almost certainly much less in error than the assumption, stimulated by 
the Laplacian theory of the origin of the earth, that our planet was fluid 
in the not very remote past. Remembering the fact that we are assuming 
simply that the apparent secular acceleration of the moon's mean motion 
is due to tidal friction and that it is a measure of the rate of tidal evolution, 
and that we are assuming further that the physical condition of the earth 
and moon has not changed in the time covered by our calculations, we find 
from (23) that when D=*20 hours the value of P was 24.096 days, and from 
(140) that P had this value 220,700,000,000 years ago. If the action of 
the sun had been included the interval would have been decreased by about 
20 per cent. It is impossible to believe that the neglected factors, such as 
the eccentricity and inclination of the moon's orbit, could reduce the time 
enough to change the order of these results. This computation, which has 
the merit of being based quantitatively on actual observations, points very 
strongly to the conclusion that tidal evolution is so slow a process that it 
can not have played an important rdle in the earth-moon system, even when 
we consider an interval of a billion years. 

There is, however, another possibility that may be considered. It is at 
least conceivable that there may be unknown forces acting upon the earth- 
moon system in such a way that they largely mask the relative secular 
tidal acceleration of the moon. Any thing increasing the moon's distance 
and period without otherwise disturbing its motion, or any thing acceler- 
ating the rotation of the earth, would tend to ofifset the seciUar acceleration 
produced by tidal friction. The possible secular contraction of the earth 
is a factor working in the right direction. But from the numbers obtained 
above it follows that, if we are to escape from the conclusion that tidal fric- 
tion is now a negligible factor, we must assume that the actual relative tidal 
acceleration of the moon is several hundreds of times V per century. Sup- 
posing the reduction to V is due to the acceleration of the rotation of the 
earth because of shrinking, it follows that at every epoch in the past the 
day and the month were more nearly equal than they would have been 
except for this factor. Finally, at the limit at which they were equal, 
their common period was many times that computed above, and their 
great initial distance fatal to the fission theory. If there are unknown 
forces retarding the moon's revolution, the conclusions are the same. 



The object of this investigation has been to examine the theory of tidal 
evolution in order to find out, if possible, not what might take place under 
certain assumed conditions, but how important this process has been in the 
actual development of our sjrstem. The aim has been to avoid, as far as 
possible, assumptions regarding the uncertain factors depending upon the 
phjrsical conditions of the bodies involved. In order to compare the theory 
with the actual facts the various methods of testing it have been carried to 
quantitative results. 

A large part of the discussion has been made to depend upon the com- 
ponents of the moment of momentum and upon the energy of the system. 
In section II the moment of momentum equations and the energy equation 
are devdoped, and they are perfectly rigorous so long as the two bodies 
are subject to no forces except their mutual attraction. Under this con- 
dition the three components of moment of momentum are rigorously 
constant, and the three equations which express these conditions are fixed 
relations among the various quantities which define the djmamical state 
of the system. The energy equation is a relation among the same quan- 
tities, but unlike the components of the moment of momentum the energy 
diminishes by friction. These relations are too few to determine the changes 
which will actually take place, but they give important information about 
them. They are particularly valuable, for they are true whatever the 
phjrsical conditions of the bodies involved. 

One of the conclusions reached by Darwin was that it is probable that 
the earth and moon have developed from an original mass by fission. One 
critical test of this hypothesis is the determination of the smallest distance 
at which the bodies could have revolved around each other consistently 
with the present moment of momentum and energy. This test has been 
worked out quantitatively, first with the problem simplified so that the 
conclusions are absolutely certain under the hypotheses; then the effects 
of various modifying conditions, which seem more or lees probable, have 
been examined, one after another, and their influence upon the final result 
determined. The results reached are so near the border line separating what 
is favorable to the theory from that which is unfavorable, that it is impor- 
tant in applying this test to determine accurately the constants upon which 
the system depends. One of these is the rotational moment of momentum 
of the earth, which depends only upon the law of density of the earth as an 
uncertain factor. In section III the constants of the density according to 
the Laplacian law are worked out. It is found that according to this law 
the density varies from 2.76 at the surface to 10.84 at the center. While this 
is probably not an exact expression for the earth's density, the inherent 
probabilities as well as the actually observed precessional phenomena lead 
us to conclude that it is not sensibly in error for the purposes of this dis- 
cussion. By the same law the surface and central densities of the moon are 
respectively 1.65 and 6.51. 

In section IV the moment of inertia for the Laplacian law of density 
is found, and it comes out 0.336 times the mass instead of 0.4 times the 
mass, as in the case of a homogeneous body. 


Section V is devoted to a consideration of the problem in which the 
two bodies revolve in circular orbits undisturbed by exterior forces, and 
in which the rotational moment of momentum and energy of one of the 
bodies are so small that they may be neglected, while the axis of rotation of 
the other is perpendicular to the plane of the orbit. Although the condi- 
tions assumed in this section are not exactly fulfilled in any physical prob- 
lem, still they are near enough those prevailing in the earth-moon system to 
throw much light on what may possibly have taken place. But the chief 
value of this investigation is that the variables are so few that the results 
are precise, except as to the time rate at which the possible changes will 
take place. The thing of greatest interest is that the rates of change of 
revolution and rotation are proportional to the rate of the loss of energy 
through friction, and are not directly dependent upon the phases and 
lags of the tides or the surface peculiarities of the tidally distorted body. 
These results show that, so far as the hypotheses upon which they are 
founded apply to the earth-moon system, if we could from the direct tidal 
observations calculate the rate of loss of energy in tides raised by the moon 
upon the earth, then we could compute the rate of tidal evolution at the 
present time. While this problem undoubtedly presents serious difiiculties, 
they are not more formidable than those of assigning to the earth a physical 
constitution which shall agree reasonably with the truth. 

Assuming that friction is proportional to the height of the tide and its 
vdocity relative to the surface of the tidally distorted body, and that the 
loss of energy is proportional to the square of the friction, equations are 
developed, (33), giving the rates of change of the periods of revolution and 
rotation. They involve only one unknown constant depending upon the 
phjrsical constitution of the distorted body. 

In section VI the equations of section V are applied to the earth- 
moon system. The influence of the sun is neglected, later computation 
showing that its effects upon the rotation of the earth are now about one- 
fifth as great as the moon's. The rotational moment of momentum and 
energy of the moon are small because of the small mass of the moon, its 
small dimensions, and its slow rotation. Using the numerical data, it is 
found that the moon's rotational moment of momentum is less thui one 
thirty-thousandth that of the earth. Since the eccentricity of the moon's 
orbit is small and the cosine of obliquity of the ecliptic not much less than 
unity, it is seen that the conditions of this investigation really approxi- 
mate rather closely to the actual earth-moon S3rstem. It is found that the 
month has always been increasing and that it can not pass beyond 47.7 
of our present days, at which period the month and day will be equal and 
the system move as a rigid body. There is no way of telling by this investi- 
gation how long a time will be required for the system to reach that state. 
But it is a more interesting fact that the month can never have been less 
than 4.93 of our present hours, this being the period of revolution when 
the distance from the center of the earth to the center of the moon was 
0,104 miles. Consequently we must suppose that when the moon broke 
off from the earth it was at this distance from it, or 5,236 miles from its 
present surface. Or, including the radius of the moon and supposing that 


both the earth and moon were of the same density and shape as at present, 
the distance from the surface of one body to the surface of the other was 
immediately after fission 4,155 miles. Since this result is altogether incom- 
patible with the obvious implications of the fission theory, we must either 
abandon the theory or show that this number would be very largely reduced 
by including the effects of the neglected factors. Consequently we examine 
the effects of various neglected conditions and influences. 

If the earth were rotating in 4.93 of our present hours it must obviously 
have been very oblate instead of spherical as was assumed in the computa- 
tion. In the absence of certain Imowledge we may assume that its equa- 
torial radius reached out to the surface of the moon when the distance of 
its center was 9,194 miles, that the oblateness was such that the volume 
was the same as at present, and that the law of density was such that its 
rotational moment of momentum was the same as it would have been if 
it were spherical and the Laplacian law of density prevailed. We find 
that under these hjrpotheses the polar radius would have been only 942 
miles. A scale drawing shows that this oblateness is out of the question, 
and a little consideration shows that the equatorial zone must have been 
so rare as to make it impossible to account for the mass of the moon. 

If we waive the condition that for a given period of rotation the law 
of density was such as to keep the rotational moment of momentum the 
same as when the body was supposed to be a sphere, we shall have to sup- 
pose the moment of momentum was greater than this in order to get sufGi- 
cient matter in the periphery to account for the origin of the moon. But 
this supposition leads to the conclusion that the nearest possible distance 
of the moon was greater than the 9,194 miles found before. 

Another hypothesis is that the earth was initially larger than at present, 
and has shrunk to its present dimensions as it cooled. It was found in this 
case also that the initial distance of the moon must have been greater than 
the 9,194 miles found on the original hypothesis. 

An examination was made of the hjrpothesis that the earth originally 
had a radius of 9,194 miles and a density such as to keep its moment of 
momentum the same as if it were of its present size. It was supposed the 
density varied according to the Laplacian law and the constants of the law 
were worked out by the conditions of the hypotheses. It turned out that 
the density of the surface must have been negative, a result having no 
phjrsical interpretation and proving the falsity of at leaist one of the hypoth- 
eses upon which the computations were made. 

In sections VII and VIII the problem was treated without neglecting 
the rotational moment of momentum and energy of the moon, but keeping 
the earth and moon spherical. It turned out that the initial distance of the 
earth and moon could not have been less than 9,200 miles. That is, when 
this factor is included the result becomes less favorable to the fission theory 
than when it was omitted. 

Then in sections X and XI the hypothesis was made that the moon's 
rotational moment of momentum and energy may be neglected, but the 
eccentricity of the moon's orbit was given the value assigned by observa- 
tions. It was found under these hypotheses that the initial distance of the 


moon could not have been lees than 9,214 miles, a result more unfavorable 
to the fission theory than that obtained when the eccentricity of the moon's 
orbit was neglected. 

In sections XII and XIII it was assumed that the rotational moment 
of momentum and energy of the moon and eccentricity of the moon's 
orbit may be neglected, but the inclination of the plane of the earth's 
equator to the plane of the moon's orbit was taken into account. Under 
these h3rpotheses it was found that the initial distance of the moon could 
not have been less than 9,364 miles, a result more unfavorable to the fisrion 
theory than any of those heretofore derived. 

All of the factors initially neglected and later taken up one by one 
have made the initial distance greater than the originally computed 9,194 
miles. Obviously all of them combined would operate in the same direction. 
Since they only increase a difficulty which was in the first place serious, it 
is not necessary to go to numerical results for all of them combined. 

The factors which remain to be considered in attempting to test the 
fission theory by computing the initial distance of the moon are the sun's 
perturbations of the moon's orbit and its effect upon the rotation of the 
earth. The first part of section XIII is devoted to a discussion of the direct 
action of the sun upon the moon's orbit, and it is shown there that includ- 
ing this influence alone the initial distance of the moon could not have been 
less than 9,206 miles, which is somewhat greater than that found when the 
sun's action was neglected. 

The second part of section XIII treats the relative retardative effects 
of the Sim upon the rotation of the earth. The magnitude of the tide- 
raising force of the sun compared to that of the moon can easily be com- 
puted. Other things being equal there are good grounds for assuming that 
the rate of tidal evolution is proportional to the square of the tide-raising 
force. The friction depends also upon the speeds of the tidal waves with 
respect to the earth's surface. At the present time the speeds of the moon's 
and the sim's tides are about equal, but if we trace the system back until 
the month and day were approximately equal this relation is no longer 
approximately verified. We must resort, therefore, to some specific assump- 
tion as to the way in which tidal friction depends upon the speeds of the 
tides over the surface of the earth. The assumption was made that friction 
is proportional to the first power of the velocity, and therefore that the 
loss of energy is proportional to the second power of the velocity. It is 
practically certain that this assumption will give results which are sensibly 
true. Using this hypothesis and supposing that the rotational moment 
of momentum and energy of the moon, the eccentricity of the moon's 
orbit, the inclination of the plane of the earth's equator to the plane of 
the moon's orbit, and the direct action of the sun on the moon's orbit may 
all be neglected, it was found that the initial distance of the moon was 
reduced from 9,194 miles to 9,045 miles. Thus it is seen that the one 
factor which makes the moon's initial distance less than that found in the 
first computation is not only of no particular consequence, but also that 
it is less than some of the factors which increase it. Using all those factors 
whose^effects have been computed when they have been supposed to act 

•epsrateljTr u>d suppCNBUig that they would be eBtentially tbe same when 
•eting jointlr, we find ^t tbe ■mallest poaiible distance of the moon 
eomimtible with present conditions is 9,241 mites. Fig. 16 sbows the earth 
and moon and their initial distance to scale, and obviously one would need 
azttemaly strong confirmatory evidence to convince him that the moon had 
just broken off from the earth. The distance from the surface of the earth 
to the surface of the moon is 4,201 miles, or 243 miles greater than the 
radius of the earth. 

As a conoesrion to the theory, we may assume that the earth and moon 
have separated by fission so that their periods of rotation and revolution 
are precisely equal, and then inquire whether the present system could 
develop from it. If the original orbit were exactly circular the orbit would 
always remain circular. Since the moon's orl^t now has considerable 
eccentricity it follows that we must assume that the orbit immediatdy 
after separation waa somewhat eccentric. But since the rotations would 
be sensibly uniform while the revolution would be such as to fulfil the law 

of areas, there would be relative motion of the various parts and therefore 
tidal evcdution. Tbe question whether this friction would drive the moon 
farther from the earth or bring it back and precipitate it again upon the 
earth is treated in section X, and it is found there, under the assumption 
that the loss of energy is proportional to the square of the tide-raising 
force and the square of the velocity of the tide along the surface of the 
earth, that the tide* vmdd bring tlie moon again to lAs earth. Thus, unless 
some of the neglected factors can offset this result, the direct implications 
of the theory destroy it, and it may be noted here that these remarks apply 
with equal force to the hypothesis that the binary stars have originated 
by fission and that their present distances from each other and the eooan- 
tridties of their orbits are a residt of tidal friction. 

If we neglect the rotational moment of momentum and energy of the 
moon, the eccentricity of tbe moon's orbit, and the inclination of the plane of 
the earth's equator to the plane of the moon's orbit, then it is certain, as was 
shown in sections V and VI, that tidal friction will at the present time 
lengthen both the day and the month, but at such relative rates that the 
number of days in a month will decrease. Consequently if time be measured 
by the rotation of the earth the moon will continually get ahead of its place. 


as predicted by the gravitational theory neglecting tidal evolution. This 
relative gain in longitude, if established by observations^ will give us the 
measure of the rate of tidal evolution at the present time, and we ean safely 
apply it, by properly varying the factors depending upon the moon's dis- 
tance, over any interval during which the physical condition of the earth 
has been essentially as it is at present. Since the geological data go to show 
that the physical state of the earth has been about as it is now for many 
tens of millions of years, and do not give certain evidences of any radically 
different general physical conditions, we are perhaps justified in boldly 
applying results based on the present rate of gain of the moon for a very 
long interval of time. 

It is well known that a comparison of ancient and modem eclipses shows 
that the moon has an acceleration in longitude of about V per century 
which is not explained by perturbations. Let us assume that this is due 
to tidal friction and is the measure of it at the present time. At this rate 
it will take over 30,000,000 years for the moon to gain one revolution. 
Consequently we see without any computation that it must have been an 
extremely long time in the past when its period was a small fraction of its 
present period. 

The problem was treated in section XV, and it was found there that, if 
the physical condition of the earth has been essentially constant, the length 
of the day was 20 of our present hours, and of the month, 24 of our present 
days not less than 220,000,000,000 years ago. It is extremely improbable 
that the neglected factors, such as the eccentricity of the moon's orbit, 
could change these figures enough to be of any consequence. This remark- 
able result has the great merit of resting upon but few assumptions and 
in depending for its quantitative character upon the actual observations. 
If it is accepted as being correct as to its general order, it shows that tidal 
evolution has not affected the rotation of the earth much in the period 
during which the earth has heretofore been supposed to have existed even 
by those who have been most extravagant in their demands for time. 
And if one does not accept these results as to their general quantitative 
order, he faces the embarrassing problem of bringing his ideas into harmony 
with the observations. 

If tidal friction has been an important factor in the evolution of the 
earth-moon system, then presumably it has also been an important factor 
somewhere else in the universe. Certainly one would expect to find the 
theory encountering no difficulties in the case of any other planet. But 
the place where it has been applied most is in the orbits of the binary stars, 
which have been supposed to have become binaries through fission and 
to have become widely separated as a consequence of tidal evolution. The 
first difficulty is, as has been pointed out, that there must have been an 
initial eccentricity of the relative orbit, and this eccentricity would cause 
the bodies to reunite as a consequence of tidal friction. But there is another 
important difficulty, as was explained in section IX. If we suppose the 
binary to be composed of two equal suns moving just after separation 
as a rigid body, and if we waive the effects of the eccentricity for the sake 
of the argument, we find that there is a maximum distance to which the 

sumiABT. 133 

bodies ean be driyen from eaeh other as a consequence of tidal friction, 
this maadmum distance corresponding to a minimum of energy, and that 
this maadmum distance is not many times greater than the distance im- 
mediatdy after sq[>aration. In particular, when the equations were applied 
to two stars each equal in mass and dimensions to the sun at an initial 
distance between centers of 866,000 mfles, it was foimd, on the basis of 
condusiye reasoning, that the greatest distance possible as a result of tidal 
evolution would be only 1,042,400 miles. These results, which were ob- 
tained under the hypothesis that the stars suffered no shrinkage with loss 
of heat, were not radically modified when they were supposed to shrink to 
any extent whatever. The conclusion is that the widely separated binaries 
which our tdeseopes reveal to us can not have originated by fission, at least 
from masses condensed beyond the nebulous stage.^ 

In a word, the quantitative results obtained in this paper are on the whole 
stronj^y adverse to the theory that the earth and moon have developed 
by fission from an original mass, and that tidal friction has been an impor- 
tant factor in their evolution. Indeed, they are so uniformly contradictory 
to its implications as to bring it into serious question, if not to compel us 
to cease to consider it as even a possibility. 

^Tbm of coune rafen only to ipontaDeoiis finion without the aecenion of moment of 
momentum from aome outvie body. 



coirrjuBmoNS to cosmogony and the fundamental psoblems of geology 




AsBOwUe Professor of Astronomy^ UniversUy of Chicago 




In the speculations on cosmogony there are two fairly definite hypoth- 
as to the manner in which a single body may give rise to two or more 
distinct masses without the intervention of external agencies. The first, as 
outlined by Laplace, is that possibly a rotating fluid may abandon an equa- 
torial ring, which will subsequently be brought by its self-gravitation into an 
approximately spherical mass. The second, the fission theory, had its rise 
in Darwin's researches on tidal evolution, and in his speculations on the 
origin of the moon. It has found extensive application in attempts at 
explaining the great abundance of binary stars. 

The hypothesis of Laplace has the support of no observational evidence, 
unless we regard the rings of Saturn as such, and rests upon no well-elab- 
orated theory. On the contrary, there are well-known considerations of the 
moment of momentum of our system which compel us to reject it as being 
an unsatisfactory hypothesis for the explanation of the development of the 
planets. But the fission theory of Darwin, even if the origin of the moon is 
left aside as being doubtful, has strong claims for attention because of its 
immediate application to explaining the origin of spectroscopic and visual 
binaries and certain classes of variable stars. Besides, it is in a general way 
confirmed by the investigations of Maclaurin, Jacobi, Kelvin, Poincar£, and 
Darwin on the figures of equilibrium of rotating homogeneous fluids, and on 
their stabilities. In particular, considering a series of homogeneous fluid 
masses of the same density but of different rates of rotation it is shown that 
there is a continuous series of figures of stable equilibrium beginning with 
the sphere for sero rate of rotation; then, with increasing rotation, passing 
along a line of oblate spheroids until* a certain rate of rotation is reached; 
then, with decreasing rate of rotation but with increasing moment of 
momentum, branching to a series of ellipsoids with three unequal axes, and 
continuing imtil a certain elongation is reached; and finally, at this point, 
branching to a series of so-called pear-shaped figures. It has been con- 
jectured that if it were possible to follow the pear-shaped figures suflSciently 
far, it would be found that they would eventually reach a point where they 
would separate into two distinct masses. From this line of reasoning it has 
been regarded as probable that celestial masses, through loss of heat and 
consequent contraction, do break up in this way often enough to make the 
process an important one in cosmogony. 

Aside from the unanswered question as to what form the pear-shaped 
figures finally lead, there are two reasons for being cautious in accepting the 
conclusions. One is that the celestial masses are by no means homogene- 



OU8. When they have reached, or are in, that condition of steady motion of 
slow rotation postulated in the investigation, they are undoubtedly always 
strongly condensed toward their centers. The other, and probably more 
important, one is that isolated celestial masses do not change their rates of 
rotation except when they change their densities or distribution of densi- 
ties. While the theoretical discussions to which reference has been made 
regard the rate of rotation as the single variable parameter, in the actual 
case there is a corresponding change in density. The importance of not 
neglecting the latter is easily seen. 

Consider a slowly rotating homogeneous fluid having the form of a nearly 
spherical oblate spheroid. The eccentricity of a meridian section depends 


upon the quantity ^ p t where w is the angular rate of rotation, X:* the 

gravitational constant, and a the density of the mass. The eccentricity of 
the axial section increases with the increase of this function, provided it does 
not go beyond a certain maximum. Now suppose the mass contracts in such 
a way as to remain homogeneous throughout, and so that it continues to 
rotate as a solid spheroid of equilibrium. Because the moment of momentum 
of an isolated mass is constant, the contraction implies an increase in o), and 
therefore, as far as this factor alone is concerned, an increase in the oblate- 
nees of the mass. But the contraction also implies an increase in a, and 
therefore, so far as this factor alone is concerned, a decrease in the oblate- 
ness of the mass. That is, keeping the moment of momentum constant, as 
the dynamical situation requires, we find the eccentricity acted upon by 
two opposing factors. If, under the influence of these factors, the figure 
should become less oblate, the fission theory would get no support from the 
discussion; if it should get more oblate, the question is at what rate the 
mass must rotate and to what extent the contraction must proceed before 
there is a possibility of fission. This paper will be devoted to a brief discus- 
sion of these questions. 



For the applications which follow it will be necessary to review briefly 
the facts regarding the spheroidal and ellipsoidal figures of equilibrium and 



their conditions of stability. 

Maclaurin^ has shown that for very small values of 'o~W~ ^^^^^ ^^^ 

two ellipsoids of revolution which are figures of equilibrium, one of them 
being nearly spherical and the other very oblate, the limits for co =» being 
respectively the sphere and infinite plane. For greater values of this quan- 
tity, the figure corresponding to the former is more oblate and that corre- 
sponding to the latter is lees oblate. For ^ j^ =■ 0.22467 . . . the two 
figures are identical. For ^ hi > 0.22467 . . . there is no ellipsoid of 
revolution which is a figure of equilibrium. 

Jacobi has shown' that if » ^ < 0.18709 . . . there is an ellipsoid of 

three unequal axes satisfying the conditions for equilibrium. When this 
quantity is very small, the axis of rotation and one other are very short and 
nearly equal to each other, while the third is relatively very long. With 
greater values of this quantity the shorter axes are longer and the longest 

axis is shorter. For » j^ =0.18709 . . . the figure becomes an ellipsoid 

of revolution and is identical with the more nearly spherical Maclaurin 

spheroid. For o 1.2 > 0.18709 . . . the Jacobian ellipsoids of revolution 
do not exist. ^s 

In the case of the Maclaurin spheroids the relation between ^ ^^ and 

the eccentricity, e, of an axial section is given by the well-known equation^ 






It follows from these equations that for i •- we have 

and that for A « oo we have 

0(i)-O e-1 -|^-co 

> Treatise on Fluxions, Edinburgh, 1742. 
' Letter to the French Academy, 1S34. 
'Tisserand, M6canique Celeste, 2, Chap. VI. 




It 18 well known that -^j^ passes through lero but onoe between X^O and 

A— 00. Hence we may represent in figure 17 the faets so far enunciated. 
As «u starts from sero and increases, there is a series of figures of equilibrium 
starting from O and another from P, the two series coinciding and vanishing 
at the point a. 

Fio. 17. 

We may also indicate the existence of the Jacobian ellipsoids of equi- 
librium on this figure, without, however, being able to define completely 
their shape by a single point on a curve. Let us represent the eccentricities 
of the sections made by planes passing through the axis of rotation and each 
of the other two axes of the ellipsoid by abscissas in figure 17. The Jacobian 
ellipsoids branch from the Maclaurin spheroids at b. For a given value of 

o JL2 the corresponding point on the curve 6c gives the eccentricity of the 

section through the longest axis and the axis of rotation, while the cor- 
responding point on bd gives the eccentricity of the section through the 
remaining axis and the axis of rotation. These two points together com- 
pletely define the shape of the ellipsoid. 

However, we shall regard the two series of ellipsoids be and bd as dis- 
tinct, the properties indicated by a point on either of them being sufficient, 
when taken with certain equations of relation not represented on the dia- 
gram, completely to define the figure. That is, each curve of the whole 
diagram will be regarded as carrying with it a certain set of equations which 
serve to complete the definition of the shape of the figure of equilibrium 
corresponding to each of its points. Thus, OaP carries with it the equation 
which sayB that the eccentricity of every plane section through the axis is e. 



The equations associated with db state that the figure is an ellipsoid, and 
relate the eccentricity given by the point of this curve with the eccentricity 
of the other principal section by means of elliptic integrals.^ From this point 
of view the figures corresponding to points on the curve bd are quite dis- 
tinct from those on be, and it is regarded simply as an interesting fact that 
they are the same in shape and differ only by their orientation in space. For 
sufiBdently small w there are two figures of this kind. When w increases 

so that 


—0.18709 • . . they become identical with each other and 

with the Maclaurin spheroid and vanish at this point. 

Fio. 18. 

Figure 18 is a scale drawing of an axial section of the Madaiuin spheroid 
corresponding to the point b of figure 17, and is therefore the figure from 
which the Jacobian ellipsoids branch. The eccentricity of its axial section 
is 0.813, or more than twice that of Saturn. In fact, there is no known 
celestial mass condensed beyond the nebulous state which approaches the 
oblateness of this theoreticflJ figure of equilibrium. 

^ TiMennd, IMesnique Celeste, 2, Chap. VII. 



In a memoir ^ remarkable for its powerful methods and important results, 
Poincar^ has proved the existence of an infinite number of other forms of 
equilibrium. He considered the equations for equilibrium as functions of the 
parameter co. For a definite value of to, as io^o)^, they have a certain 

number of solutions. For example^ for ^ l2 < 0.22467 there is a solution 

on Oa and one on Pa of figure 17. If -o~h~ < ^- 18709 there are also solutions 

on he and hd. If for co=cO| two or more solutions unite and do not vanish 
as (o passes through a>|, then the figure of equilibrium corresponding to w^ 
is a form of bifurcation. If after uniting they vanish, the figure is a limit 
form. Thus, in figure 17, h belongs to a form of bifurcation, for at this point 
the Maclaurin spheroids and Jacobian ellipsoids are identical. The point a 
belongs to limit form, for at this point the two series of Maclaurin spheroids, 
Oa and Pa, unite and vanish. Likewise the point h belongs to a limit form 
for the series hd and he. 

Poincar6 showed in the work cited that there are no forms of bifurcation 
corresponding to points on the curve Oh, but that there is an infinite num- 
ber of them on haP. He proved also that there is an infinite number of 
them on he and hd between the point h and the axis a> =^0. That is, in addi- 
tion to the spheroids and ellipsoids of equilibrium an infinite number of 
other forms exist. The first one on he is at /, and its deviation from the 
Jacobian ellipsoid to the first order of small quantities depends upon the 
third zonal harmonic with respect to the greatest axis of the ellipsoid. It 
is the pear-shaped figure referred to above. Since it is unsymmetrical with 
respect to the axis of rotation, there are really two similar figures, differing 
by 180^ in orientation, just as the two series of Jacobian ellipsoids differ by 
90° in orientation. There is of course a precisely similar series on hd. 

If two real series of figiures of equilibrium, A and B, cross, and if before 
crossing A is stable and B unstable, then after crossing A is unstable and B 
has at least one degree less of instability. Poincar£ has also proved an en- 
tirely similar theorem in periodic solutions of the problem of three bodies.^ 

All the spheroids corresponding to points on the curve Oh of figure 17 are 
completely and secularly stable. At h the spheroids lose their stability but 
the branching Jacobian ellipsoids are stable. They remain stable until /is 
reached. It is an interesting question whether the pear-shaped figures are 
stable or unstable. Poincar^ threw the determination of the answer to the 
question into a form capable of numerical treatment,* and Darwin has 
made an elaborate and detailed discussion of it.* The rigorous answer turns 

* Sur r^uilibre d'line masse fluide animde d'lin mouvement de rotation. < Acta Mathe- 
matica, 7, 1885, 269-380. 

' Les M^thodes Nouvelles de la M^canique Celeste, 3, pp. 347-349. 

' Sur la stabilit^^ de r^uilibre des figures pyriformes affectdes par une masse fluide en 
rotation. < Phil. Trans., A, 198 (1902), pp. 333-373. 

*The stability of the pear-shaped figure of equilibrium of a rotating mass of liquid. 
<Phil. Trans., A, 200 (1903), pp. 251-314. 



on the numerical value of the sum of an infinite series; and from a compu- 
tation of its first terms Darwin regards it as certain, though not algebrai- 
cally proved, that the pear-shaped figure is stable. However, liapounoff 
has stated^ that these figures are unstable. If so, the line of completely 
stable figures terminates at this point, and as soon as a body has passed be* 
yond it a slight disturbance will cause it to undergo radical changes of form 
and perhaps break into many fragments. Even if the pearH9haped figures 
are at first stable, they may become unstable as well as all figures which 
branch from them long before fission occurs. Indeed, this now seems prob- 
able, for Darwin has found,' in a memoir on the figure and stability of a 
liquid satellite, that a satellite loses its stability before it can be brought 
near enough to its primary to coalesce with it. 

Consequently if a slightly viscous fluid mass were originally turning 
slowly and had the form of a stable Maclaurin spheroid of equilibrium, and 
if in some way greater and greater rates of rotation were gradually impressed 
upon it without violently disturbing its figure, then we should see the series 
of changes in its shape described by a point moving along the curve Ob to 
the point b, then branching on to the line be of stable forms, again branching 
at/, if the pear-shaped figures are stable, and continuing along lines of stable 
figures until they terminate or until fission takes place. At any rate there 
is no possible chance of fission until the change of shape has' passed beyond 
b, for up to this point there is secular stability and no branching. In fact, 
we may feel assured that it can not occur until the shape of the mass has 
passed at least to/. In order that we may see to what a remarkable extent 
a rotating homogeneous fluid must depart from sphericity before there is a 
possibility of fission starting, we give in figure 19 the most oblate section 

Fio. 19. 

of the ellipsoid belonging to the point/. Darwin has shown' that its eccen- 
tricity is e» 0.9386. The eccentricity of the other principal section through 
the axis of rotation is e' = 0.6021, and the eccentricity of the principal section 
perpendicular to the axis of rotation is 6^=^0.9018. 

* Aoad. Imp. dee Sd. de St. P^rsbourg, 17, No. 3 (1905). 

* Phil. Trans., A, 206 (1906), pp. 161-248. 

' On the pear^haped figure orequilibrium of a rotating mass of liquid. < Phil. Trans., A, 
198 (1902), pp. 301-331. 


The points of bifurcation and conditions of stability or instability depend 
only upon the shape of the figures, while the shapes of the figures in the va- 

rious series depend only upon the values of ^ ^ . Consequently if a rotat- 
ing homogeneous fluid body contracts in such a way as to remain always 
homogeneous, its shape and condition as to stability are determined by this 
function throughout the whole series of changes. 





We nofw enter a fidd bent with fonnidaUe difficulties and in which 
there aie but few p iM iti fe leeaUe. We shall consider masses which are 
posaesBed off dow rotation, whieh grow continually denser toward their 
eentersy and wlueh are a p projumatdbr qpherieal in form. To simplify the 
pioblon we may mpp oee first that we have under consideration a body 
composed off a number off tnesaiprBsnUe fluids of different densities, arranged 
in order off increanig density from thesurfaoe to thecenter. This hypoth- 
esis more neady agrees with the conditions found in nature than that of 
homogeneity does» and the results obtained under it may be taken as throw- 
ing h^t on the aetnal problems. 

Oairant has diown* that such a body as we are considering will always 
be less oblate than it would if its mass were uniformly distributed through- 
out its Tdume. This is easy to see in conadering the limiting case of a 
dense nudeus surrounded by a homogeneous atmosphere of vanishing mass. 
That this result may be kgitimatdy applied to the cdestial bodies is proved 
by the fact that the earth, Jupiter, and Saturn are all less oblate thim they 
would be if they were homogeneous and rotating at their respective rates. 
It is an interesting and important fact that the differences in oUateness 
of these planets and the eoneqponding homogeneous figures of equilibrium 
are greater the smaller the mean density. That is, if a low mean density 
means the mass is largdy gaseous and compressible, we may condude that 
the more a body b condensed toward its center the less oblate it will be for 
a pvea rate of rotation. The f aets for the earth, Jupiter, and Saturn are 
given in the following taUe, whoie e has been computed from equation (1). 

eeecuiiicttjr of 









Condder still the ease of a heterogeneous incompresdble fluid mass with 
greatest dendty at its center. Suppose the dendty is given by an equa- 
tion of the form /o\ 

where a^ is the mean dendty of the whole mass, and where a^ is dways 
fiidte and may be a continuous function of c. For t^O certain series of 
figures of equilibrium are represented in figure 17. Let us suppose there is a 
third axis in the figure perpendicular to the e and lo-axes. We shall mark 
off vdues of c along this axis. 

Let the conditions for a figure of equilibrium be 


x^,w,t)^0, 1-1, 


See TfaMraad's MtauBiqoe Cfleste, 2, Chap. Xm. 



For €»0 and ^ ^ < 0.22467 . . . certain solutions are represented by 

points on the curves of figure 17. Suppose for w^w^ the solution x^^^x^^^^ 
is not a multiple solution. The F^ depend upon the gravitational potential 
and the rotational energy, and are continuous functions of t. Consequently 
the roots of (3) vary continuously with t, and we may represent their solu- 
tion by a point in figure 17 in the plane e«£, where as before, each linear 

series with respect to a parameter, as ^ u or e, carries with it a set of 

relations which completely defines the shape of the figure of equilibrium. 
If <o=a>o and e«0 belong to a multiple solution of (3) there is an a> near 
io^ such that for e— e equations (3) have also a multiple solution. By this 
process the curves of figure 17 become surfaces to ^every point of which 
belongs a figure of equilibrium. If in this figure e is set equal to a small 
constant a new set of curves will be obtained in a general way similar to 
the old, and possessing maxima and points of bifurcation. In general the 
greater i the greater will be their deviation from the forms of the curves 
in figure 17. 

Now consider the question of stability. The necessary and suflicient 
condition for complete and secular stability of the figure of equilibrium 
is that the total energy shall be a minimum for all variations preserving 
constant moment of momentum.' All the quantities involved in these 
conditions are continuous fimctions of c. Consequently, starting from an 
ordinary point in figure 17 whose corresponding figure of equilibrium has 
any properties of stability, it is found that the figure obtained by varjring e 
through a sufficiently small range will have the same properties of stability. 
Consider the curves obtained by giving £ a constant value. At certain places 
the figures of equilibrium will change the character of their stability; but as 
in the case of c = 0, treated by Poincar^, wherever the stability changes a 
new series of figures branches out. Since the curves for e=e are in the 
analytic sense the continuation of those for e = 0, the figures of equilibrium 
for €=£ go through a series of changes of stability entirely analogous to the 
changes in the figures for e = 0. Of course, it is possible that two curves, 
Co' and Co^, might cross a curve Co at a single point for e-^O, and that the 
corresponding curves, G/ and C/, might cross the curve C. in two distinct 
points, or the opposite. For example, there might be such a definition of a^ 
that, for a certain value of e, the point corresponding to / of figure 17 would 
fall on the point corresponding to 6. However, in the present connection 
such exceptional cases are trivial. The point of interest is that for € = e>0 
there is a line of stable figures of equilibrium corresponding to those for 
which this parameter is zero. 

In general, for e > 0, the point of bifurcation corresponding to 6, figure 
17, will not appear for the same value of to as that belonging to b. We may 
represent these two values of lo by a>. and u)^ respectively. The question 
of interest in the present connection is which is the greater; or, in other 
terms, whether with increasing rotation instability occurs first in the het- 
erogeneous or in the homogeneous body. We shall not attempt a positive 

> Thomson and Tait's Natural Phil., Part II, 778, (/) and ijk). 


answer to this question, but in view of the fact that for a given rotation 
and mean density the homogeneous body is more oblate than the hetero- 
geneous body, we are justified in concluding that probably instability first 
appears in the figures of equilibrium for e « 0. 

Now consider the case of a heterogeneous compressible rotating fluid. 
The preceding remarks pertaining to the existence of the figures of equi- 
librium still holdi though all the explicit defining equations are immensely 
more complicated. But the question of stability has a new element in it. 
As Jeans has shown in an important memoir/ gravitation itself becomes a 
source of instability. It is easy to see that if for any reason there is a local 
condensation in a compressible fluid, gravitation will tend to augment this 
condensation. On the other hand, the elastic forces called into play tend to 
destroy the condensation. The two kinds of forces are in conflict, and the 
state of stability depends upon which will predominate when the equilibrium 
is disturbed. Jeans showed that an infinitely extended nebulai of density 
such that it possesses the properties of gases, is in unstable equilibrium 
independently of its mean temperature and density. That is, in this case it 
is possible to introduce such local variations in density that the decrease in 
gravitational potential energy shall more than balance the increase in the 
potential energy of the elastic forces. 

There are as yet no quantitative results by means of which we may 
measure the importance of this factor of instability. It is true that Jeans' 
and Love * have proved the stability of the earth under certain assumptions, 
but in view of its present existence, notwithstanding the many vicissitudes 
which it has survived, this result may be taken as reflecting favorably upon 
the method employed in obtaining it, rather than assuring us of the perma- 
nency of our planet. In the absence of positive quantitative results we are 
able to make only more or less probable hypotheses as to the importance 
of this factor. 

There are seen to be two opposing factors entering into the question of 
stability of heterogeneous compressible masses. As compared with homo- 
geneous incompressible masses, the central condensation tends strongly 
toward sphericity, as is shown both by theory and by the observed shape of 
the sun and planets, and therefore presumably toward stability. But the 
compressibility tends toward instability for local deformations and altera- 
tions in density. We shall assume as appearing reasonable that in bodies 
having strong central condensations and continuous changes in density these 
two opposing factors approximately balance. To the extent to which this 
assumption is justified, we may draw conclusions in regard to the actual 
celestial bodies from what is known regarding the forms and stabilities of 
homogeneous fluids. For the purposes of additional safety in this procedure 
we shall keep far from the conditions of possible disruption in the applica- 
tions which follow. 

* Tbe stability of a spherical nebula. < Phil. Trans., A, 190 (1902), pp. 1-53. 

* On the vibrations and stability of a gravitating planet. <Phil. Trans., A, 201 (1903), 
pp. 167-184. 

* The gravitational stability of the earth. <PhiI. Trans., A, 207 (1908), pp. 171-241. 





Equation (1) is the relation connecting the shape of the Maclaurin 
spheroid; its density, and its rate of rotation. Let m represent its mass, a 
its polar radius, and M its moment of momentum. Then we have 

m=|;ra«(l+;i')a (4) 

J|f=|ma'(l+;i> (5) 

Eliminating cj and a between equations (1), (4), and (5), we have 

For a homogeneous body of given mass and moment of momentum, this is 
a relation between the density and oblateness which must always be satis- 

fied so long as the figure is a spheroid. It is easily verified that -r^ ifl 

positive for all real values of X, from which it follows that when the left 
member of the equation is given there is but a single real solution for X\ 

The Jacobian ellipsoids are defined by the equations^ 

/^ V(l-C^(l-AM^»COdC (7) 

/ (H->l'C*)»(l+>l'*C*)» 



where the axes of the ellipBoid are a, b, and c, and 

6»-o»(l+>l*) c»-a»(l+il'*) 

Equation (7) defines the relation between X and X' which must be satisfied 

by these figures of equilibrium, and equation (8) expresses — in terms of 
X* and X'*. " 

The equations corresponding to (4) and (5) are in this case 

m-|jra»'y/T+^-y/T+A^ff Jlf =^(2+^>+^'»)<^ (9) 

By means of these relations, equation (8) reduces to 

25(|;r)«M» , (2+ A 

SJfc'mV " " (l+A') 

Now we may write (7) and (10) respectively 

^'+^'') [ ^ c'a-C)dc (10) 

*(1+>1")* / (l+>l'c')*(l+>l"C')* 

0{X,X')=O ,piX,X')=Kai (11) 

* Tisserand's M^anique Celeste, 2, Chap. VII. 



where X is a constant depending upon m and M. From these equations we 



Then we find 


\ K dd 

Zd (tI dk 


aii'- + 

1 K d«. 






r, Since i and X' enter and f symmetrically, it follows that when the 
ellipsoid branches from the Maclauiin spheroid, t.e., when X^H, we have 

■TJ-— rw-. Hence as the ellipsoid branches from the spheroid because of 

increasing denmty, the eccentricity of one principal axial section increases 
while that of the other decreases. This continues indefinitely unless either 

^ or Tw vanishesi which is extremely unlikely. This means the figure 

tends to become cigarnshaped. At a certain elongation (see figure 19) the 
so-called pearnshaped figure branches. Certainly in homogeneous masses 
there can be no fission before this elongation, with its corresponding den- 
sity, is attained. 



In applying the formulas above to the solar syBtem we must remember 
that they are strictly valid only when the masses are homogeneous. Now 
the sun and planets are certainly not homogeneous, but we have seen reasons 
for believing that nevertheless the formulas will give results which are not 
remote from the truth. But, because of this uncertainty, in the applica- 
tions which follow we shall not attempt to draw conclusions except where 
the margin of safety is extremely great. 

Let us consider first the sun. We shall find its density for various de- 
grees of oblateness, and its oblateness for a certain very high density.* Re- 
ferring to (6) we see that the greater the moment of momentum of a body 
the less dense it will be for a given oblateness, and the more oblate it will be 
for a given density. Consequently we shall be favoring the conclusion 
that the sun will eventually suffer fission if we use too large a value of M. 
The moment of momentum is most easily computed if we suppose the sun 
is homogeneous, and the result obtained in this way will certainly be in 
excess of the true value. 

Using the mean solar day, the mean distance from the earth to the sun, 
and the mass of the sun, as the units of time, distance, and mass respec- 
tively, we find that the density of water is 

<^wAter'= 1,567,500. 

Taking the sun's density as 1.41 on the water standard, its period of rota- 
tion as 25.3 days, and its radius as 433,000 miles, we find for its moment of 
momentum oia 


Now we may apply equation (6) to find how dense the sun will be before 
the Jacobian ellipsoids branch off. It is hardly possible that the sun could 
suffer fission before this point is reached, the shape of the spheroid being 
given in figure 19. The computation shows that when the sun shall have 
reached this degree of oblateness its density will be 

<T=307 X 10" on the water standard. 

This density corresponds to an equatorial radius of the sun of 11 miles. 
Since this density is millions of times greater than it is supposed matter 
ever attains under any circumstances, we must conclude that the oblate- 
ness of the sun can never approach that for which the Jacobian figures of 
equilibrium branch. Or, in brief, the sun can never contract so much that 
its rotation will threaten it with disruption. 

Notwithstanding the extreme character of these figures, one might still 
be so ultra-skeptical as to doubt the conclusion, since it is based on the com- 
putation of the point of bifurcation for homogeneous masses. However 
that may be, we must admit that the sun will be stable until its oblateness 
reaches that of Saturn at present. We have seen that the eccentricity of a 

' Strictly speaking the computations are made for homogeneous bodies of the same 
mass and having the same moment of momentum, but no confusion will result from this 
mode of expression. 


meridian section of a homogeneous fluid having the mass and the mean 
density of Saturn, and rotating in its period, would be 0.607. Since the 
constitution of the sun is in a general way like that of Saturn, it will have 
the oblateness of Saturn about when the section of the corresponding 
homogeneous mass has this eccentricity. Making the computation from 
equation (6) we find that the sun will not become so oblate as Saturn is now 
until its mean density becomes 148 X 10^ on the water standard. This cor- 
responds to a radius of 37.3 miles. If we may regard this as an impossibly 
large density we may conclude that the sun will never be so oblate as Saturn 
is now, and that its stability will always be greater than that of Saturn at 

Apparently the chances that Saturn will separate into two parts because 
of shrinking and rapid rotation are greater than that any other member of 
the solar system will ever suffer fission. To examine the probabilities we 
shall apply equations (4), (5), and (6) to Saturn. Taking the density, mass, 
and period of rotation as 0.72, j-^f fti^d 10.25 hours respectively, and com- 
puting the moment of momentum under the hypothesis that Saturn is now 
a homogeneous sphere, in order to give the theory that fission is possible all 
the benefits of the approximations, we find that when Saturn shall have an 
oblateness equal to that of the spheroids from which the Jacobian ellipsoids 
branch, its density will be 21 times that of water, its axial diameter 16,500 
miles, its equatorial diameter 28,400 miles, and its period of rotation 1 hr. 
24 m. The high mean density demanded seems to be fatal to the theory of 
fission in this case. 

In order to see how great changes in the density, dimensions, and period 
the body will undergo by the time it reaches the state where the pear- 
shaped figures branch, we may apply equations (7), (9), and (10) to Saturn. 
Darwin has made the computations^ from equations equivalent to (7) and 
(8), and has found that for this point 

X = 0.7544 X' - 2.7206 k^^ — 0. 1420 

Then equations (9) show that at this stage the mean density of Saturn 
must be 93 on the water standard, its polar diameter 9,400 miles, its longest 
*diameter 27,000 miles, and its period of rotation 46 minutes. That is, the 
mass is about four and one-half times as dense when the pear-shaped figures 
branch as when the Jacobian ellipsoids branch. While the computation 
was applied to Saturn, it follows from equations (6) and (10) that this same 
ratio for the densities at these critical forms is true, whatever the mass and 
moment of momentum of the body under consideration. 

The density which the earth will attain before it will reach one of the 
critical forms is so great that the computation is without interest. But we 
may examine the hypothesis that the earth and moon were originally joined 
in one mass whose rapid rotation produced instability, and that resulting 
fission gave rise to two bodies having great stability. It is to be observed 
in the first place that the moment of momentum of the earth-moon system 
has remained constant except for influences exterior to itself. There is none 

' Phil. Trans., A, 198 (1902), p. 326. 


readily assignable which could have increased it. Among those which may 
have decreased it apparently the solar tidal friction is the only one which 
can have produced sensible results. 

The effect of the moon's tides in the earth has been to transfer moment 
of momentum from the earth to the moon. According to the results given 
in the preceding paper on tidal evolution, section XIV, the maximum 
moment of momentum the earth-moon system could have had, including all 
that the sun could have taken from it, is that belonging to the system mov- 
ing as a rigid mass with a period of 4.8 hrs. Let P represent this common 
period. Then, neglecting the inclinations of the planes of the equators of 
the earth and moon to the plane of their orbit and the possible eccentricity 
of their orbit, factors which reduce the moment of momentum, we have by 
equation (6), loc. ciL, 

The quantities c^ and c, depend upon the distribution of mass through- 
out m^ and m,, m^ representing the earth, and m^ the moon. If the masses 
are homogeneous c^^c^^QA, If they obey the Laplacian law of density 
Ci»C2« 0.336. If the distribution of mass is such that the densities increase 
from the surfaces to the centers of these bodies, the values of c^ and c, are 
less than 0.4. We shall certainly get too large a value for M by putting 
C|=>C2»0.4. Adopting these values we find that for the earth-moon system, 
in the units of tWs paper, Af =8X10~". 

Since large moment of momentum tends to instability we shall favor the 
theory of fission if we add 26 per cent to this number, supposing that per- 
haps this amount may have been lost through meteoric or other friction. 
Then, using Af =10~" and m = 3 X 10~*, equation (6) gives for this united 
earth-moon mass at the time when the Jacobian ellipsoid branched a mean 
density 215 times that of water. We find similarly that this hypothetical 
earth-moon mass could not become even so oblate as Saturn is now imtU 
its density had become 10.4 times that of water. Since the present density 
of the earth is only 5.53, this means that if the hypotheses upon which this 
computation was made are valid, the earth-moon system can not have arisen 
from the fission of a parent mass under the influence of rapid rotation. 

In the preceding paper, starting with the earth-moon system as it now 
exists and following backward in time the effects of tidal friction, it was not 
possible to get the earth and moon in close enough proximity to make the 
fission theory seem possible. Now, starting with the supposed initial sys- 
tem with the critical factor, the moment of momentum, determined from 
observations, we do not find the figure approaching an unstable form until 
the density is more than 40 times the present density of the earth. 



Direct telescopic observations prove the existence of very many binary 
stars. According to Hussey and Aitken about one star in 18 of those 
brighter than the ninth magnitude is a visual double. There are many 
stars whose spectra periodically consist of double lines. This phenomenon 
is taken as indicating that these stars are binary systems, made up of 
two approximately equal components, whose orbital planes pass nearly 
through the earth. There are many more stars whose spectral lines period- 
ically shift. This is interpreted as meaning that these systems are binaries 
in each of which one component is relatively non-luminous. There are many 
variable stars whose light curves are explained by supposing they are eclips- 
ing binary systems. Unless the interpretations of these phenomena are 
very much in error a considerable fraction of all the stars are binaries. 
Granting that the interpretations are correct, the evolution of binary sjrs- 
tems is a standard celestial process. We are raising the question here 
whether these binaries may have originated from parent masses by fission 
without external disturbing factors, and if so at what stage of condensation. 

Before considering the question of the fission of single masses into binary 
systems, we shall write down some of the implications of the interpreta- 
tions of the spectra of spectroscopic binaries, particularly as regards limits 
on their masses and densities. 

Consider first the case where both spectra are visible. Let i be the com- 
plement of the inclination of the plane of the orbit to the line of sight. 
Suppose the orbit is circular. Let v be the maximum observed radial 
velocity, roi and m, the two masses, P the period of revolution, and a the 
major semi-axis of the orbit. 

Then we have 

^^* k^[^^^, 2;rt^cos»t ^' ) 

Since i can not be determined in a star which is not also a visual binary we 
have, reducing the units so that when v is expressed in kilometers per second 
and P in mean solar days, the sum m^+m^ expressed in terms of the sun's 
mass by the relation ...^ 

m^+nH^-^Pv^ (16) 

Suppose the spectrum of only one star, m^, is visible, and let v^ be its 
maximum observed radial velocity with respect to that of the center of 
gravity of the system. Let r^ be the radius of its orbit, assumed to be cir- 
cular, around the center of gravity of the system. Then we have 

Pr, o . D 27rrii(mi+m,) 

— V=27rr, wiir. — m/-, a^r^+r^ P— — ^ , — — 

cost * II rj 11 j.^1 



whence '^^i V ^'^J (1«) 

' 2wJk* cos* t 


In this case the lower limit of the mass of the system can not be computed 
except an assumption be made regarding the relation between m^ and m,. 

Suppose — -=fi> Then equation (16) gives 

m,+m,= (l+/i)m,>^??J^±^* (17) 

When P is expressed in mean solar days, and v^ in kilometers per second, we 

m^ + nh^^Pv,\l+piy (18) 

where m| + m, is expressed in terms of the sun's mass. If it were true that 
the more massive star always gives the observed spectrum we should have 
/«>land ..24 

^i+^>-^^V (19) 

Let Oi and a, represent the radii of m^ and m, respectively. Let the dis- 
tance between their surfaces be represented by #ca. Then 0^#c< 1 and we 

a.+a,-a(l-«) P= 2;r(a.+aOt ^^^ 

A;(l— ic)*-ym,4-m. 

Suppose the two bodies have the same density a. Then we have from 
m^^-K^noa^t m^^-^aa^ and equation (20), putting — ^=/i as before 

Zn (1 +/!*)* . 

The ratio /< may vary from zero to infinity, and x from to 1. From the 
derivative a<; 3;r (i+;zi)^(l-;z») 

d;£ Jk2p2(l-K)» /iJ(l+;£)2 

it follows that; for fixed values of P and k, a constantly increases while ii 
varies from to 1, and then constantly decreases while /i varies from 1 to 
00 . Therefore, since (21) is a reciprocal equation in /i, we have 

^^'^ ^<^^ , .,./!., o. (22) 

Or, changing the units so that a will be expressed in terms of the density of 
water when P is expressed in mean solar days, we have 

* ^^'^TT^TwA:^. (23) 

100 (1 - kYP^ ^ - 100 (1 -i«)»F 

The smaller k the smaller the limits for a. When the bodies are in contact, 
and the period 4.57 hours, that of ^ Cephei,^ we find 

» The Period of f^ Cephci, by E. B. Frost, Astrophysical Jour., 24 (1906), pp. 269-262. 


If they are separated so that their surfaces are at a distance from each 
other equal to one-half the sum of their radii, i.e., it k^^, equation (23) 

According to Darwin's results, loc. eit., p. 232, this is about the minimum 
distance at which homogeneous masses could revolve in stable equilibrium. 
If they were separated farther the already high limits on the density would 
be still greater. It follows that we must conclude either that these very 
short period binaries and eclipsing variables are very dense, or that hetero- 
geneous masses are more stable than homogeneous ones. 

Now let us return to the question of fission of binary stars. Denoting 
the periods of rotation of m^ and m, by D^ and D, respectively, we have for 
the total moment of momentum of the system ^ 

(2^)»(m,H-ng»"^ D, ^ D, ""^^^ 

The signs of the second and third terms in this equation are determined 
under the hypothesis that both bodies rotate in the direction in which they 
revolve. But this is a necessary consequence of the fission theory, and 
therefore an allowable assumption in testing it. 

If P > D| and if P > D, then the mutual tides of the two bodies tend to 
bring P, D^ and D, eventually to the same value. In the case of widely 
separated visual binaries the fission theory implies that the tidal evolution 
has proceeded far, and that Dj and D, are closely approaching an equality 
with P. If these conditions are satisfied and if m^ and m, are approximately 
equal, we see from (24) that the inequality 

^>(2^7*W,+m,)* ^^^^ 

is nearly an equality. For example, in the case of a Centauri, assuming that 
each component has the dimensions of the sun and that D| » D, « P approx- 
imately, the ratio of the first term to the sum of the other two is roughly 
10,000 to 1. In general, the greater P and the more nearly equal m^ and 
m,, the more nearly the inequality (25) approaches an equality, and the 

Let us suppose the two stars were originally in one spheroidal mass 
m^m^-\-m^. Then equation (6) gives the relation between its density and 
oblateness, which reduces by means of (25) to 

25m,»m,»fctPI > ^ (1 H-^')« f (3 +k') \ .gO) 

6(3^)*(m,H-m,)^^<-^— j-T-**^ ^"^j ^^^^ 

Now let ZT'^l^) then equation (26) gives 

^<6(3^/fVa:^|(3+£)t,„_.,_3| (27) 

Letting /(/i)-- ^^-^^ we find for /i-l that -1^-0, 4-t>0 and that ^' 

d/£ ' dp? dft 

* See equation (6), p. 85. 


vanishes for no other value of ft. Consequently the right member of (27) 
has a single minimum at /i = 1. But for other values of fi the inequality of 
(27) differs more from an equality, and hence we are not certain that /c= 1 
gives the least value of a for a given P and A. But in many stars ft is 
undoubtedly near enough unity to make equation (27) useful. 

We shall suppose pL = l and compute o for the value of X for which ap- 
parently there is first any danger of fission. If we suppose that this first 
occurs for that value of X for which the Jacobian ellipsoids branch, that is 
for X=' 1.395 . . . , we find, taking the units so that a will be expressed in 
terms of the density of water when P is expressed in terms of mean solar 

^*^®' 016 

a<^ (28) 

Or, if the fission occurred when the pearnshaped figures branched from the 
Jacobian ellipsoid, we find similarly 

^ 0.071 ™. 

Consequently, this discussion leads to the conclusion that in all binary sys- 
tems in which the two masses are approximaidy equal, and in which the periods 
are at least several years, as they are in the vistud pairs, ti^e fission mtut have 
occurred, if at all, while the parent mass was yet in the nebulous state. The 
data regarding binary systems as a class are so meager that probably no 
stronger conclusion than this can be drawn from this line of argument. 

There is, of course, no a priori objection to the theory that binaries as a 
class have originated by fission in the nebulous state. But there are at least 
two rather distinct hypotheses as to how and why such fission may have 
taken place. The first is that in the origin of a nebulous mass the factors 
which have determined its initial condition may have brought it into exist- 
ence with at least two nuclei of condensation whose magnitude and density 
were sufficient to have led to a binary, even though the moment of momen- 
tum may have been so low that, if the mass had been spheroidal with the 
same mean density, it would have been a stable figure. Fission in this t3rpe 
of masses is not under consideration here. The second is that the mass in 
its earliest nebulous stage was in an approximately spheroidal form, densest 
at its center with density decreasing outward through approximately sphe- 
roidal layers, and that as a consequence of its high moment of momentum 
it lost its stability and divided into two masses. This is the type of fission 
under consideration here. 

Suppose a nebula of this latter type suffers fission. At the time of fission 
all parts are rotating at the same angular rate, and one of the two parts 
must have a mean density less than, or at the most equal to, the mean den- 
sity of the original mass. Consequently one of the two fragments because of 
its lower density and equal rotation, must have at least as great a tendency 
to fission as that which led to the division of the initial mass, unless cither 
its form is one of greater stability, or the tidal forces of the other member of 
the pair tend to keep it from breaking up. If, as seems probable, the ap- 
proximate spheroid is the most stable figure of equilibrium, and if the mass 


under consideration has suffered fission by evolution along this line of figures, 
as is assumed, then the former alternative is eliminated. It does not seem 
that the tidal factor can tend toward stability. This intuition is strongly 
supported by the results obtained by Darwin, loc, cit., attacking the problem 
from the other end, viz., that it is not possible to bring two homogeneous 
fitiid masses near enough to touch without their being certainly in unstable 

We observe next that the binary stars are now actual stars of consider- 
able density. Consequently if they have originated from the fission of neb- 
ulas they have undergone enormous contraction. The contraction implies 
increased rotation which would increase the already dangerous tendency 
for at least one part to suffer further fission. Tidal friction would offset 
this tendency by decreasing the rotations, but considering all the factors 
involved, it is seen that if a fitiid mass ever gets to the state where fission 
occurs, there is at least great danger of its breaking into many pieces. 

Consequently we are led to believe that if binaries and multiple stars 
of several members have developed from nebulas, the nebulas must orig- 
inally have had well-defined nuclei. The photographs of many nebulas 
support this conclusion. But we observe that if we are forced to this 
position we do not explain anything — we only push by an assumption 
the problem of explaining the binary systems a little farther back into 
the unknown. 



The problem under consideration is that of the fission of celestial bodies 
because of rapid rotation when they are not disturbed by important exter- 
nal forces. The attack is made through well-known results concerning the 
figures of equilibrium and conditions as to stability of rotating homogeneous 
incompressible fluids. It is recalled that for slow rotation a nearly spherical 
oblate spheroid is a stable form of equilibrium; that for greater rates of 
rotation the corresponding figure is more oblate; that when the eccentricity 
of a meridian section becomes 0.813 the figure loses its stability and at this 
point a stable line of three axis ellipsoids branches; that when the longest 
axis of the ellipsoid becomes about three times the axis of rotation a new 
series, known as the pear-shaped figures (or better, perhaps, the cucumber- 
shaped figures) branches, and that before this point is reached there is no 
possibility of fission. We are almost entirely ignorant as to what may happen 
after this point is passed, and it must be remembered that it has not been 
proved that in any case fission into two stable bodies is possible. 

The celestial bodies differ from those just considered in two important 
respects. In the first place their densities increase toward their centers. 
For a given rate of rotation and mean density this central condensation 
makes them more nearly spherical, as is shown both by theory and by com- 
parison of the observed figures of the planets with the computed forms of 
corresponding homogeneous masses. In the case of Saturn, for example, 
the eccentricity computed on the hypothesis of homogeneity is 0.607 while 
the observed value is only 0.409. It seems certain that this central conden- 
sation tends toward stability. The second important difference between 
the ideal homogeneous incompressible fluids and the celestial bodies is that 
the latter are compressible. This latter factor, at least under certain cir- 
cumstances, tends toward instability. 

The opposing quantitative effects of central density and compressibility 
undoubtedly differ greatly in different masses and can not be easily deter- 
mined in any case. However, if we may assume that they approximately 
offset each other, we may reach some conclusion respecting the possibility 
of the fission of the actual celestial bodies by discussing the corresponding 
homogeneous incompressible body. This is the assumption adopted here, 
but, because of its uncertainty, in the applications to the solar system, where 
it turns out fission is impossible, all approximations are made so as to favor 
fission, and it is assumed that in the actual bodies fission may be immanent 
long before it is possible in the homogeneous ones. These safeguards and 
simplifications are possible and easy because it is a negative result which is 

The actual problem is not one in which the rate of rotation changes 
while the density remains constant, though this is the one heretofore treated 
in the mathematical discussion. In the physical problem the rate of rota- 
tion and the density change simultaneously with the shape in such a way 
that the moment of momentum remains constant. Imposing this condition, 
we arrive in the case of the spheroids and ellipsoids at relations between 
the density and respective shapes, the coefficients depending upon the mass 


and the moment of momentum. When the oblateness of the spheroid is 
given there is but a single density satisfying the conditions, and when the 
density is given there is but one spheroid satisfying the conditions. 

For the applications we assume that an actual celestial body will not be 
in danger of fission until the corresponding homogeneous incompressible 
body arrives at the state where the Jacobian ellipsoids branch. The density 
at this stage is less than one-fourth that at wUch the pear-shaped figures 
branch, and actual fission in the homogeneous bodies is certainly beyond 
this form, if indeed fission into only two bodies is ever possible. With this 
very conservative assumption we proceed to some calculations. 

(1) We find that the sun can not arrive at this critical stage until its 
mean density shall have exceeded 307 X 10" on the water standard. This 
corresponds to an equatorial diameter of the sim of about 22 miles. 

(2) We find that the sun can not become so oblate as Saturn is now until 
its mean density shall have exceeded 148 X 10^® on the water standard. This 
corresponds to an equatorial diameter of the sim of about 75 miles. 

Since even the latter density is impossibly great we conclude that the 
sun will never become so oblate as Saturn is now, and that it will always be 
more stable than Saturn is now. 

(3) We find that Saturn can not arrive at the critical stage at which the 
Jacobian ellipsoids branch until its mean density shall have become 21 
times that of water. This corresponds to a polar diameter of 16,500 miles 
and an equatorial diameter of 28,400 miles. We conclude because of the 
great density demanded that Saturn will never su£fer fission. 

(4) We assume that the earth and moon were once one mass and get 
their original moment of momentum from its present value. In computing 
it, however, we make certain approximations so as to get it too large and 
thus favor the conclusion of fission, then we add[to it the maximum amount 
the sun's tides can have taken from the earth, and finally we add 25 per cent 
for fear there may be some unknown sensible factors omitted. Then we 
find that this hjrpothetical earth-moon mass could not get even to the 
critical point where the Jacobian ellipsoids branch until its mean density 
became 215 times that of water, or about 40 times the present mean density 
of the earth and moon. It would not become even so oblate as Saturn is 
now until its density had become 10.4 times that of water. Therefore we 
conclude that the hypothetical case was false, and that the moon has not 
originated by fission from the earth in this way. 

(5) In applications to the binary stars the results are less definite because 
of the meager data regarding these systems. But assuming that fission in 
stars will occur when the Jacobian ellipsoids branch in the corresponding 
homogeneous masses, we find for the density a in terms of water at the time 
of fission when the two stars are of equal mass 

^ 0.016 

where P must be expressed in mean solar days. Even though fission should 
not occur until the density is ten times this amount (which, if true, makes 
the evidence against fission in the solar system much stronger), all visual 


binaries of two approximately equal masses must have separated, if they 
have originated by fission, while they were yet in sr nebulous state. The 
results are of the same order so long as the disparity in the two masses 
of a binary is not very great, and this probably includes all of the visual 

(6) Certain formulas, not connected with the question of fission, were 
developed for binary syistems. If P represents the period in mean solar 
days, fi the ratio of the mass of the star whose spectrum is measured to the 
mass of the other one, and v^ the maximum observed radial velocity ex- 
pressed in kilometers per second, then the sum of the masses expressed in 
terms of the sun's mass must satisfy the relation 

When the spectra of both stars are measurable, and v represents the 
maximum relative velocity of the stars, the corresponding formula is 

If we let a represent the distance between the centers of a binary pair, 
and ML the distance between their surfaces, and suppose they have the same 
mean density a, then a must satisfy the inequalities 



100(1 -/c)«P» = = 100(1 -/c)«P» 

where a will be expressed in terms of water when P is expressed in mean 
solar days. 

The results obtained by the computations above are quite adverse to the 
fission theory, in general, except if it is applied to masses in the nebulous 
state, and seem practically conclusive against it so far as the solar system 
is concerned, either in the future or past. Perhaps the hypothesis that 
stars are simply condensed nebulas, which has been stimulated by a cen- 
tury of belief in the Laplacian theory, should now be accepted with much 
greater reserve than formerly. Up to the present we have made it the 
basis not only for work in dynamical cosmogony but also in classifying the 
stars. It may be the time is ripe for a serious attempt to see if the oppo- 
site hypothesis of the disintegration of matter — because of enormous sub- 
atomic energies, which perhaps are released in the extremes of temperature 
and pressure existing in the interior of sims, and of its dispersion in space 
along coronal streamers or otherwise — can not be made to satisfy equally 
well all known phenomena. The existence of such a definitely formulated 
hypothesis would have a very salutary effect in the interpretation of the 
results of astronomical observations. We should then more readily reach 
what is probably a more nearly correct conclusion, viz., that both aggre- 
gation and dbpersion of matter under certain conditions are important 
modes of evolution, and that possibly together they lead in some way to 
approximate cycles of an extent in time and space so far not contemplated. 





ProfenoT of Oeology, Unhenity of Chicago 



It is a familiar view that a rotating spheroid of gas may, by cooling and 
shrinking, so far accelerate its rate of rotation as to cause its own sepa- 
ration into two or more parts. The resulting parts are assigned various 
relative values, and the separated masses are given different forms, ranging 
from fragments and rings to subequal masses. This view of possible self- 
partition has found expression in various cosmogonic conceptions from the 
nebular hjrpothesis to the formation of binary stars. 

To consider the bearings of molecular activity in a representative case, 
let a spheroid of gas be chosen whose mass is comparable with that of the 
solar system and whose volume is such as may be hjrpothetically assigned 
it. Let its rate of rotation at the outset be such that the value of gravita- 
tion at the equatorial surface is greater than the centrifugal component of 
rotation. Let cooling follow, in consequence of which the rate of rotation 
will be progressively accelerated. Let it be assumed — as has usually been 
done — ^that the rate of rotation would at length reach such a velocity that 
separation in some form would take place regardless of any question as to 
the manner of its realisation. Our question relates to the effect of mo- 
lecular activity on the transition from an undivided spheroid to a spheroid 
divided in some way, whether by massive fission, into larger or smaller 
fractions, or by individual molecules. 

In a body whose molecules are boimd together into a coherent mass, 
such parts as may be affected by like general stresses are properly treated 
as units, within the limits of cohesion, but in a body whose molecules possess 
all degrees of freedom, and which act with complete individuality, the treat- 
ment may with special appropriateness be based on the molecule as the 
unit. Molecular action in a gaseous spheroid consists of encounters or 
quasi-encounters, and of rebounds or quasi-rebounds along free paths be- 
tween the encounters. Within the mass, the excursions and encounters of 
the molecules give rise to an effect equivalent to viscosity which influences 
the movement of one part of the gaseous mass upon another part, and may 
perhaps have given rise to an impression of coherence; but on the outer 
border of the mass — the critical portion in this case — ^this effect becomes a 
vanishing quantity, and individuality of action is dominant. 

According to the laws of gaseous distribution, the density of a gaseous 
spheroid, when controlled solely by its own gravitation, declines from a 
maximum at the center progressively toward the surface, where the limit 
of gaseous tenuity is reached and an ultra-gaseous state supervenes. The 
transition from the gaseous to the ultra-gaseous state is the critical factor 
in the case, since it is at the extreme surface of the gaseous mass that the 
11 163 


centrifugal component of rotation first comes into equality with the cen- 
tripetal force of gravitation and constitutes a condition precedent to the 
separation of any mass of gas as a body. 

In the depths of the gaseous spheroid the paths between encounters 
may be assumed to be relatively short and hence straight, since gravitation 
can not sensibly affect paths of brief duration. At higher levels, the free 
paths grow progressively longer, and at length horizons may be reached at 
which the attenuation permits free paths of such length and duration that 
they may be appreciably curved by the gravitation of the spheroid. At 
still greater heights the attenuation reaches such a degree that curved 
paths come to dominate and, at a certain stage of rarity, a portion of the 
molecules rebounding from encounters in outward directions, find no 
molecules in their paths, and therefore hold on their coiu'ses until arrested 
and turned back by gravitation, if its force be sufficient, or else they pass 
on beyond the limit of the spheroid's control. Theoretically, under 
the Boltzman-Maxwell law of molecular distribution, a certain small per- 
centage of molecules should reach the parabolic velocity of the spheroid 
and escape, but for the purposes of the present discussion this fraction need 
not be considered independently of a larger class to be described presently, 
with which it may be merged as having like influence on the moment of 
momentum of the spheroid. 

Such of the outward-bounding molecules as are arrested by the sphe- 
roid's gravitation obviously turn back toward the spheroid without a re- 
versing encounter and thus describe elliptical loops. In this they differ 
markedly from the molecules in the depths of the gaseous spheroid, whose 
paths are sensibly straight and whose courses are terminated by encounters 
at either end.^ In the elliptical courses the outward movement is ter- 
minated by a gradual decline in the molecule's speed until its outward 
progress is reduced to zero, when there follows a new movement inward 
accelerated by gravitation. 

If the to-and-fro, collisional activity of molecules constitutes the essen- 
tial characteristic of a gas, the outer border of the strictly gaseous part of 
the spheroid should be placed at the transition zone where the molecules 
cease to-and-fro passages between encounters and begin to describe ellip- 
tical loops limited outward by gravitation, but this demarcation is rather 
a matter of convenience than an essential in the consideration of the modes 
of action. 

In the course of their outgoing and incoming movements, the molecules 
pursuing elliptical paths are subject to collision with one another. The 
phases of such encounters may vary indefinitely and the velocities of the 
rebounding molecules may represent an indefinite variety of interchanges 
of kinetic energies. Inspection shows that some of these molecules must 
rebound toward the gaseous spheroid, that some must take distinctively 
new elliptical paths, while a certain proportion will inevitably be thrown into 
courses more or less tangential to the surface of the spheroid, and some of these 
may have sujfficient velocities to assume orbits aboiU the spheroid, and thus form 

^ This distinction has been drawn by G. Johnston Stoney, Astrophys. Jour., vol. XI, 
1900, pp. 251 and 325, and elsewhere. 


a revolvtional system of molecules dynamically, independent of the spheroid, 
except that they act as miniUe satdlites. 

In a stationary spheroid the rebounds that give rise to revolutional 
courses are as likely to take one direction as another, and if the mass of the 
spheroid be great, the number of molecules which will acquire revolution 
from molecular activity alone may be neglected in this discussion. 

But in a spheroid in a state of rapid rotation, especially a spheroid 
approaching the critical stage of centrifugal separation, the molecules shot 
outwards in the direction of rotation will start with the sum of the common 
velocity of rotation and the individual velocities acquired from the last 
encounter, while the molecules shot in a direction opposite to the rotation 
will have only the difference between the common velocity of rotation and 
the velocity acquired from the last encounter, the meridional component 
in each case being neglected as immaterial here. It follows from this that 
when the velocity of rotation is high, the molecules starting from encoun- 
ters in the direction of the spheroid's rotation will much more largely pass 
into orbital paths than molecules starting in the opposite direction. 

In a spheroid having the mass of the solar system and a radius equal to 
the radius of Neptune's orbit, the equatorial velocity required for separa- 
tion by mass is above 5 kilometers per second, while the average molecular 
velocity of all known molecules, at a temperature of 2000^ C. and standard 
terrestrial pressure, falls below this. The average molecular velocity of 
most known substances falls much below this even at 4000^ C. It seems 
clear therefore that, for most of the known molecules, the effect of molecu- 
lar velocity directed backward is merely to destroy a part of their rotational 
speed, and that they still move forward relative to the center of the sphe- 
roid. With a spheroid having the solar mass and a radius equal to the dis- 
tance of the earth from the sun, and hence a separation-speed of nearly 30 
kilometers per second, only a very small fraction of the molecules could 
acquire velocities sufficient to neutralize their rotational velocities at the 
critical stage of separation. The number of molecules that could acquire 
the 60 kilometers per second required to neutralize their rotational veloci- 
ties and add sufficient velocity to give them an orbital course in a retro- 
grade direction must obviously be negligibly small in a case of this kind. 
Practically all molecules must be regarded as having forward courses with 
velocities which are either enhanced by being shot forward or retarded by 
being shot backward. 

The velocity of centrifugal separation is practically identical with the 
velocity of circular revolution about the spheroid in a minimum orbit. 
Larger orbits involve lower velocities but require additional potential en- 
ergy and moment of momentum. When the rate of rotation of the spheroid 
is very near, or essentially at, the critical stage of centrifugal separation, a 
slight addition to the velocity of an outer molecule in a forward direction, 
arising from molecular interaction, will give to it a velocity greater than 
that required for the minimum circular revolution; and before the critical 
state has been actually reached, all molecules on the equatorial periphery 
which receive forward impulses of any appreciable amount will have more 
than the requisite velocity for minimum circular revolution. If all mole- 


cules whose projections have forward components are regarded as fulfillmg 
these conditions, nearly half the molecules given outward projections will 
be included. 

Encounters and rebounds in gases under familiar terrestrial conditions 
range into the billions per second, but encounters are much less frequent 
in rare gases. Excluding outward flights that do not properly belong to the 
true gaseous state, the maximum period between encounters is less than 
the rotation-period of the spheroid, and the average period is much leas 
than that. Since interchanges are thus frequent, all molecules are liable to 
receive a forward impulse within a brief period. 

Molecules which simply receive a forward and outward projection 
greater than that requisite for free circular revolution about the spheroid 
do not, however, enter upon free orbits, directly, in most cases, because the 
paths on which they enter, the orbital in type, normally lead back to their 
starting-points, and in nearly all cases they cut the spheroid before they 
return to these points and thus complete a free orbit. If this were univer- 
sally and inevitably true, the way to free orbits along this line of evolution 
would be effectually barred. There are three lines of escape from this 
result, the first and second of which are probably unimportant; the third 
is probably effective. 

1. The first is the case in which molecules receive impulses from molec- 
ular interaction in lines tangent to the points of impact, and hence take 
elliptical paths about the spheroid which return tangentially to the points 
of impact as their peri-spheroidal climax. Their liability to encounter the 
spheroid is thus limited to these tangential touches which, in the rare con- 
dition of the gas at these vanishing points of gaseous organization, will 
not necessarily involve capture. Such molecules will not, however, be free 
from collision with the molecules pursuing elliptical loops above the gaseous 

2. By hypothesis, the spheroid is shrinking, and if the rate of shrinkage 
is appreciable during the free flight of molecules whose paths only slightly 
cut the surface of the spheroid such shrinkage may leave these paths free, 
so far as the gaseous spheroid is concerned. 

3. The two cases just named are perhaps more serviceable in defining 
conditions where gradations rather than sharp limits prevail than as 
sources of free orbital paths. The most important case is built upon the 
action of the ultra-gaseous molecules outside the gaseous spheroid, as lim- 
ited above. These molecules start from the outer part of the spheroid— 
strictly, from all depths from which there is an open path outward in the 
line of their projection — and pursue elliptical courses with return to the 
spheroid, except in the cases just noted. We have seen that, in the rep- 
resentative case of the solar system, the rotational velocity is so great, 
relative to the average molecular velocity, that most of the molecules will 
pursue forward courses, even when directed backward, and hence will be 
moving in harmonious directions. Considered as independent molecules, 
they constitute a corona of particles rising in curves, predominantly at low 
angles, and descending at similar angles to the spheroid. They are liable 
to collide in these courses and a certain percentage of collisions is inevi- 


table. Collisions in the rising parts of the curved courses take precedence 
in time, and hence in probability, over collisions in the declining courses; 
for if collision is realized in the first part of the course the molecule is likely 
to lose its chance in the latter part by being either thrown back to the sphe- 
roid by the first collision or else thrown outwards where collisions are less 
imminent. In any case, the collisions probably result either in an earlier 
return of the molecules to the spheroid, or in throwing them into new 
paths, of the orbital type, which will bri'ng them back to this point of last 
collision and not to the spheroid. This point of collision lies above the sphe- 
roid, and does not require the orbit to cut any part of the spheroid, though 
it may do so in a portion of the cases. The predominant effect will appar- 
ently be to drive the outer molecules into larger orbits and throw the inner 
ones back to the spheroid. Apparently this will be a self-adjusting process, 
so far as frequency and efficiency are concerned, for the number of molecular 
flights per unit of time will be cumulative as the acceleration of rotation 
approaches the critical stage when, as we have seen, any molecular incre- 
ment forward will lead to quasi-orbital ffight. This will increase the 
contingencies of collision, and hence a cumulative number of molecules 
will be driven into independent orbits. 

Now the most significant element in this process is the partition of mo- 
ment of momentum that is involved. Each molecule that passes into a 
free orbit necessarily takes with it more than a mean portion of moment of 
momentum. Those molecules which make elliptical ffights and return to 
the spheroid without collision carry back whatever moment of momentum 
they took out, but those thrown into permanent orbits retain, as a rule, 
not only what they took out but also the additional moment of momentum 
gained from the collisions which gave these free orbits. It follows that every 
molecule that goes into a free orbit takes a disproportionate amount of the 
moment of momentum of the spheroid and thus reduces its rotation, or else 
retards its increase of rotation, to that extent. 

If the quantitative value of this loss of moment of momentum by the 
spheroid could be compared with the increment of rotation assignable to 
shrinkage, it would be possible to determine whether the spheroid could 
ever, tmder these conditions, reach the critical stage requisite for the sep- 
aration of any portion of its mass bodily. A mode by which a rigorous 
demonstration can be reached has not yet been found, but, from the nature 
of the case, I entertain, with others, the view that the separation must take 
place molecule by molecule, and it seems to me inevitable that these mole- 
cules must go into orbits each carrying an excess of moment of momentum 
at the expense of the spheroid, and hence that the critical stage of exact bal- 
ance between the centrifugal and centripetal factors of the spheroid is never 
reached. If so, bodily separation is excluded by the conditions of the case. 

The conviction that such rotating gaseous spheroids must shed portions 
of their matter molecule by molecule, if they do so at all, has long been 
held by students of the subject, but I am not aware that the loss of moment 
of momentum from the spheroid has been urged as a reason why the crit- 
ical state prerequisite to bodily separation may not be attainable. 





Iruiructor in AppHed McUhemaHa 




This paper is devoted mainly to a quantitative study of that portion of 
the earth's internal energy which is supposed to have been derived from the 
mechanical energy of a primitive system of planetesimals, of its transforma- 
tion into thermal form during the epoch of accretion, and its subsequent 
redistribution by conduction. 

In Part I a theory initiated by Fisher is developed on the basis of the La- 
placian law of density, together with certain auxiliary assumptions. Form- 
ulas and tables are given showing the variation of dimensions and internal 
densities of the mass during the epoch of accretion, the differential effect of 
deposit of a stratum on the size and moment of inertia of the mass, and the 
deformation of mass-elements accompanying the resulting compression. De- 
termination, under alternative secondary postulates, of the original distri- 
bution of temperature produced by the compression and its redistribution 
by conduction shows the existence of a characteristic zone of rising temper- 
atures during the earlier stages. 

Part II comprises an inquiry as to what changes in the results of Part 
I are produced by changes in the secondary hypotheses employed and a 
critical examination of the latter. The computed masses of the nucleus at 
various stages of accretion are compared with the observed masses of the 
smaller planets in the solar system. The previous theory is reviewed, with 
the substitution of Roche's formula for the density; and to serve as basis 
of comparison, certain other laws of density are deduced to satisfy special 

Criticism in the light of general thermodynamics leads to a recognition 
of the theory given as possibly an extreme view, referring to a substance 
where the work of compression is mainly frictional. 

In Part III is outlined a contrasting theory for the case of a substance 
such that the work of compression is done mainly against volume-elasticity, 
under the assumption that the successive strata deposited at the surface are 
reduced to uniform entropy by free radiation while exposed. The thermal 
phenomena in this case are compared with those under the conditions of 




The following studies were undertaken at the suggestion of Prof. T. C. 
Chamberlin, as auxiliary to the development of the hypothesis put forward 
by him as to the origin of the earth by planetesimal accretion, a main 
object being to secure quantitative inferences which might aid in forming % 
judgment regarding the probable efficiency of thermal energy, whose source 
is gravitational, as an agent of geological importance under the restrictions 
imposed by the hypothesis. Experimental evidence regarding the behavior 
of substances under the enormous temperatures and pressures met with in 
the interior of cosmic bodies must be considered almost wholly lacking, and 
derived by highly uncertain extrapolation from determinations made within 
the limited range accessible to laboratory measurement. Precise conclusions 
based on accurate observed data could therefore not be looked for, but 
it was felt that it should be possible to deduce with some confidence at least 
the order of magnitude and general features of the thermal phenomena of 
gravitational origin, under the conditions assumed by the hypothesiB in 
question, when supplemented by certain minor hjrpotheses. 

The general hypothesis assumes that the earth, in common with other 
bodies of the solar system, was formed by the accretion of planetesimil 
masses, more or less similar in chemical composition, at least when consid- 
ered on a large scale, so that the more important local differences in the body 
of the resulting planet are to be ascribed to differences in physical condition, 
chiefly in pressure and temperature; and that the history of the earth in 
this aspect comprises two main epochs, the earlier one of growth by accretion, 
first at a rapid, and later at a declining rate, shading into the subsequent 
longer period of relative quiescence and constancy of mass, accompanied 
by a gradual redistribution and partial loss of a store of thermal energy 
derived from the primitive mechanical energies of the system. 

According to the mode of transformation this energy may be treated as 
mainly of three kinds: (1) that which is stored in the underlying mass, 
through the progressive static compression which accompanies the deposi- 
tion of the successive layers at the surface; (2) that derived directly from 
the kinetic energy of the masses impinging on the surface, through the vis- 
cous damping of waves and vibrations due to the impacts; (3) that derived 
in a similar way from motions which arise from the continual disturbance 
of equilibrium produced by the surface accretions, independently of the 
momentum of impact. 

To trace out exactly the final distribution of the second and third kinds 
would be a matter of forbidding difficulty, even if an acceptable assumption 
could be made regarding the precise law of accretion. No attempt will be 
made here to account for the third, but from estimates made in the sequel, 
where attention is confined to the first two, it would appear to be a rather 
small portion of the whole, being in fact under a certain special set of con- 
ditions strictly zero. 

As to the second kind, however, the energy of impact, a useful estimate 
can be made very simply if it be assumed that the single ideal substance 
contemplated is highly viscous. For in such cases the motions produced by 


an impact do not travel far from their point of origin before being practi- 
cally wiped out by frictioni thus exhibiting a kind of mechanical radiation 
of the energy brought in kinetic form by the impinging mass, but in such a 
way that f rictional absorption confines the distribution of that energy to the 
neighborhood of its source. This sort of process would be closely imitated 
if the surface structure were that of a loose aggregation of small masses, 
even if the latter were perfectly solid. It is clear, therefore, that after the 
transformation the impinging mass retains only a small part of its own orig- 
inal energy, but it secures a certain compensation from the masses whose 
deposition occurs in time and place near its own. Now, when the average 
or normal energy of impact per unit-mass does not vary sensibly during the 
time required for depositing a layer whose thickness is somewhat in excess 
of the radius of influence implied above, then the compensation may be 
regarded as practically exact, except as affected by direct radiation into 
space during the time that the mass remains exposed. 

With this interpretation it seems a fair equivalent to assume that each 
planetesimal mass retains its own primitive kinetic energy after impact in 
thermal form, but immediately loses a portion by ordinary radiation before 
it is covered up. This setting of the matter will be accepted hereafter, and 
it will be further assumed that the process of accretion, though slow enough 
to permit the loss of a large portion of the heat of impact by immediate 
radiation, is yet sufficiently rapid so that internal conduction has not time 
to modify sensibly the distribution of heat arising from compression before 
the growth is complete. 

It may also be supposed that in connection with high viscosity the mass 
would possess sufficient plasticity to enable its own gravitation to keep it in 
a condition approaching hydrostatic equilibrium, with an approximately 
spherical form, aside from the secondary effects of rotation and consequent 
polar flattening; for it is supposed that in the long run the accretion would 
be practically equable over the whole surface and that the effects of tem- 
porary inequalities of serious magnitude would be quickly obliterated. 

It is well understood, by analogy with the behavior of such materials as 
wax and pitch, that the combination of plasticity and viscosity, such as 
here contemplated as appearing under slow changes, is in no way inconsis- 
tent with the appearance of extreme rigidity under the action of sudden 
or rapidly varying forces. It should be noted, however, that a satisfactory 
theory as to the history of the earth's dominant surface features seems to 
require that to the earth-substance be attributed a rigidity sufficient to allow 
the alternate accumulation and subsidence of shearing strains, deep in the 
body of the earth, to such an extent that the periods involved, though short 
perhaps in comparison with the durations implied in phenomena of thermal 
conduction in bodies of cosmic size, are nevertheless of higher order than 
the periods of precession, nutation, and tidal phenomena, which have 
hitherto furnished the chief data pointing to the practically perfect extreme 
rigidity of the rotating earth. The hypothesis of practical fluidity tmder 
slow deformation must therefore be understood only as a crude first approx- 
imation from a geological point of view. But on account of its simplicity, 
and because of its occurrence in previous theories of the earth's constitution. 


it is desired to develop first the consequences of this assumption, leaying for 
later study the question of the modifications needed to allow for rigidity. 

The following developments refer entirely to the simple ideal case of s 
spherical body, radially symmetric in every essential feature, originating by 
deposition of successive spherical layers, and maintained under its own grav- 
itation in hydrostatic equilibrium which is at least approximate during the 
period of growth and practically exact thereafter. All variables represent- 
ing the physical magnitudes concerned are therefore considered as functions 
of the time and of the distance from the center of the mass. It is desired 
to study the primitive distribution of thermal energy due to compression 
and impact, together with the character and rapidity of the modificatioDB 
brought about by conduction and radiation. 

The term " thermal energy " used in the foregoing refers to the entire in- 
trinsic energy of the substance as depending on pressure, density, and tem- 
perature; part being the stored or latent energy of the compressional strain, 
the remainder appearing in a corresponding augmentation of temperature. 
It is the latter portion only which is subject to direct transfer by pure con- 
duction, though its redistribution in that way, through alteration of the 
geometric distribution of the mass by thermal expansion and contraction, 
may lead to the redistribution also of the energy of strain, accompanied in 
general by further transformation of energy from gravitational to thermal 
form. The character of the phenomena might easily vary radically with 
variations in the relative importance of these two portions of the intrinsie 
energy, without inconsistency with the general hypothesis. 

It is thus essential, for the construction of a definite theory, to include 
further assumptions as to the thermodynamic properties of the earth-sub- 
stance, which should cover three main points: (a) the characteristic equa- 
tion of the substance or relation between the thermodynamic coordinates—- 
pressure, density, temperature; (b) the form of the intrinsic energy as a 
function of these coordinates; (c) the value of the thermal conductivity in 
terms of the same variables. Auxiliary coefficients such as specific heats, 
thermal expansion, and compressibility can then be deduced and the as- 
sumptions checked or numerical parameters determined by means of obser- 
vations or estimates of these physical magnitudes in the case of substances 
at the earth's surface. It is evident, from the number of these secondary 
hypotheses needed, that any sharply crucial test of the main hypothesis 
from the present point of view is out of the question; all that can be done 
is to form a judgment as to its plausibility in accounting for the play of 
thermal and gravitational forces in geologic history, by developing several 
alternative suppositions on these secondary points. 



The following notations will be used throughout, and special values of 
the variable referring to the center and surface of the earth denoted by sub- 
scripts 0, 1, respectively: 

t»time. ^»los8 of potential energy in cou- 
rts distance from center. traction from infinity. 
x^rjr^. B = total energy of compression. 
/t> = density. c= energy of compression per unit- 
p= pressure. mass, 
m » mass within radius r. J^ mechanical equivalent. 
A; = constant of gravitation. JJ» modulus of cubic compression. 
g = acceleration of gravity. a » specific heat. 
V = gravitation potential. 8 = temperature. 

^ = conductivity. 

Numerical values are given, unless otherwise stated, in terms of the units 
centimeter, gram, second, and centigrade degree, for convenience in using 
published data on the absolute values of the physical constants; and are 
based chiefly on the following assumed constants, unless expressly stated: 

Total radius r^ = 6.370 X IV mean density p^ = 5.516 

surface gravity g^—QSl 

From these are obtained the following: 

volume = 1.083 X 10" total mass m^ = 5.972 X 10*^ * = 6.665 X 10"^ 

Further are assumed : 

^1=2.70 to 2.75 (estimated average for surface-rock) 
J = 4.2 X 10' ff J = about 4 X 10" a^ = about 0.2 X^ = about 0.005 

As stated above, it is necessary to supplement the general hjrpothesis by 
certain assumptions as to the physical properties of the earth-substance, in 
particular the form of the characteristic thermodynamic surface: 

F(p,^,(?)-0 (1) 

and the form of the intrinsic energy, conveniently an expression of the type: 

eHiP^O) (2) 

The fundamental equations forming the basis of the theory may then be 
grouped in two classes, related to two curves, not necessarily identical, on 
the surface (1). The first curve belongs to the actual distribution of the 
physical magnitudes within the earth at a given time, its projection in the 
P'p plane being determined in parameter form by the values of p and p as 
functions of r. The second curve corresponds to the path of compression 
traversed by a particular mass of the substance from the time of its deposi- 
tion to the time when the growth of the planet is complete, and is practi- 
cally an adiabatic curve if the accretion and, consequent compression are 
relatively rapid, as is here supposed. 


In the first class come the equations: 



-^ gp withp,-0 (3) 

km ,-. 

m— 4;r / pr^dr (^^ 

which express the condition of hydrostatic equilibrium and yield the im- 
portant relation: 

subject to the conditions 

P.-0 , %)-0 

To these must be added the expression for the potential energy ex- 
hausted during contraction from infinity: 

2 / " / "" 


+4ff / prdr W 

the latter being equivalent to 


Fo=47r / prdr or V^ 


u = F-7i (10) 

then equations (9), (3) give 

^(,«^)+4.^.-0 (11) 

where p may be considered expressed in terms of u or r, since u is necessarily 
and p most probably a monotonic function of r in the concrete case. 

The preceding equations then suffice to determine in terms of r all vari- 
ables involved, if supplemented by a single hypothetical equation, such as 
an expression for p in terms of r or a relation between p and p. 


The second class of equations, in addition to (1), (2), includes the equa- 
tions of the particular path of compression in question, together with the 
expression for the work done in compressing unit-mass: 

e- / -S.dp (12) 

which gives through integration by parts the relations 

ep+p —/OS d{ep) "sdp (13) 

the useful auxiliary variable 8 being defined by 

P (H) 

where the elastic bulk-modulus H is 

The limits of the integrals correspond to the supposition that the mat- 
ter is compressed from surface density and zero pressure. With e so deter- 
mined, and expressed in terms of r by means of equations of the first class, 
the total energy of compression is 

B= / edm-47r / epr'dr ^ / 8f^^dr (^6) 

the last form being obtained by integration by parts and equation (13). It 
may be noticed that the quantities ku and 8 are of the same physical dimen- 
sions, but with distinct theoretical setting; in case, however, the two paths 
on the thermodynamic surface are identical, they are equal at every point 
in the body. In this special case the value of E may be given as 

An essential feature of the present hypothesis is the necessity of suppos- 
ing that most of the energy of impact is wasted by immediate radiation, so 
that the compressional energy whose total is E plays the main part in the 
succeeding phenomena. It is therefore important for purposes of compar- 
ison to determine what ratio the quantity E bears to the total energy 

An idea on this point may be obtained by considering the case of a planet 
formed by condensation of a primitive homogeneous sphere of density p^ 
Suppose that a particle at distance r from the center in the completed planet 
lay momentarily at distance / when deposited at the temporary surface, 
and in the homogeneous sphere would lie at distance fj^ these being subject 
to the inequalities rj^>r'>r. It may be considered that the particle fell 


from distance fj^ to distance r*^ striking the surface, and then being covered 
up more and more deeply, settled finally to distance r. The diagram gives 
an idea of the distribution of density at the instant of deposition of that 
particle, A being the center, the part CD along the radius referring to the 

I ^ i 


homogeneous spherical shell which is being drawn up to supply surface 
deposits at JB, BC the empty region through which the particles are falling, 
and AB the radius of the partially formed planet, the variation of density 
within which is determined by the condition of equilibrium under a definite 
law of compressibility. 

Since the mass already deposited is m, the energy of impact per unit- 
mass is under these special conditions: 


(k-K) <•" 

where m, /, Tj^ are to be thought of as functions of r; then the total energy 
of impact is 

E.^^nk I pm(^~~-)r'dr (18) 

The quantities E and £| are portions of the potential energy # — #j^ ex- 
hausted during condensation of the homogeneous sphere to a condition of 
density matching that of the completed planet, ^^^ being the value of ^ for 
a sphere of assigned mass m^, and uniform density p^, and determined by 

♦.-l*^ rA-^ (19) 

The remainder of the energy ^ —^h, if any, is to be treated as of the third 
kind named above; but in case the equilibrium in the mass already depos- 
ited at each moment of the process is adiabatic, so that the two thermo- 
dynamic curves mentioned are identical, it seems probable that this third 
kind does not exist, and the total amount ^ - ^j^ transformed is completely 
accounted for as £+£^. This is later proved to be the case for one partic- 
ular pressure-density law. In any case it is of course not meant that the 
primitive distribution need be at all like that in the homogeneous sphere 
mentioned, but any other supposition would modify only the energy of im- 
pact, leaving the above point of view still useful as a check on the compu- 
tation of the compressional portion E; this is verified by the expression of 
£i as the difference 




the first member being the energy of impact corresponding to parabolic 
velocities, the second being identical with ^j^. The former would corre- 
spond to formation from a Cfystem of small particles each of density p^t but 
infinitely dispersed, so that the potential energy available is <P. It should 
be noted also that the part E is not affected by the existence of velocities 
in the primitive planetesimals due to attractions other than that of the 
nucleus on which they fall, and can therefore be treated independently 
of any supposition as to their distribution and motions previous to the 




A definite form for the development of the theory has been initiated by 
Fisher/ on the basis essentially of three main suppositions as to the prop- 
erties of the ideal earth-substance. These are: (1) that the path of com- 
pression traversed by any particular element of the mass is identical as far 
as it goes with that defined by the relation between density and pressure 
within the earth in its final state; (2) that pressure and density are related 
as specified by the classic law of Laplace; (3) that rise of temperature dur- 
ing compression is proportional to increase of compressional energy, or of 
work done to produce compression. 

For critical purposes it will be necessary to undertake a close scrutiny of 
these assumptions, as to their agreement with pertinent observed data, and 
also, in the light of general thermodynamic laws, of their consistency with 
each other, or at least of the exact interpretation to be accorded them in order 
to assure consistency. But in view of their close affiliation with the stand- 
points adopted in many previous studies of the constitution of the earth's 
interior, it will be worth while to develop their consequences in some detail. 

The pressure density law, proposed by Laplace,' has been used by many 
writers on geophysics, partly it would seem on account of its mathematical 
convenience, being the only one which reduces equation (6) to linear form. 
It is based on the condition 

-^^hp A=const. (20) 

which, when applied to the compression of an individual portion of the 
mass, gives 

P=|0>'-/>.*) (21) 

and for equations (12) and (14) the particular forms 

. = *^' (22) 

2 p 

s-hip-^p,) (23) 

Assumption (1), however, which gives s=A;u, allows the use of the relation 
(20) in the differential equation (6) or (11), which reduces to 

the appropriate solution being 

sin n r^^^ 

P=Po-j^ P=9r » (25) 

* Rev. O. Fisher. On Rival Theories of Cosmogony. Am. J. of Sc, xi, 1901, p. 414. 

* Laplace, M^canique Celeste, book xi, chap. iv. 



leaving the constants p^^, qtohe adjusted according to observation and thus 
determine h indirectly. 

Formulas for the principal variables are, then/ 

m - ^ (sin p-p cos p) (26) 

47rfc|t?o sin j8— ff cos P .^^^ 

9-— ^ (27) 


^{(¥)"-(¥)'} <»' 


^-f+f 0>-ft) or 7-ip(«i^-co8ft) (29) 

For comparison of the variables p, p, e, with their values at the center, there 
are the ratios 

p sinff 

Po" P 




^P ^ / /p-M 

while the mean density is 

p^^^Zp^mhu^^lA (33) 

or in form of power series: 

^, ft»V 31 6! "^ 7! / 

The expressions for E and 4 take particular forms readily reduced to 


giving for these total energy values: 

S.*?^V.{l+^'-4(?^»)'+3^«cos/?,| (34) 

' Most of these fonnulas occur in Fisher's paper, or in antecedent wiitingi where the 
Laplacian law is used. 



The constants p^, q, may be determined; for example, so as to give 
accordance with any assumed values for mean density p^ and surface den- 
sity p^, or the latter may give place to the condition resulting from obser- 
vations of precession and polar flattening,^ which, according to somewhat 
uncertain theory, indicate for the angle P^ a value in the neighborhood of 
140^. The following computations are based on the assumed value 141.8^, 
together with the numerical constants as listed in the introduction, giving 
the following table under the assumed law: 




3.886 X 10-* 
3.620 X10» 
2.466 X10*» 

The resulting value of the surface dendty pi, though perhaps rather 
Bmall, accords fairly well with eslimates of the mean density of superficial 
strata, and an additional check on the applicability of the density-law in 
question is found, as pointed out by Fisher, in the fact that the surface 
value H^ of the elastic bulk-modulus ranges close to values found by direct 

The accompanying tables, then, show the distribution of the principal 
magnitudes at the close of the epoch of aggregation. The columns referring 
top,p,g are of course not novel, but are inserted for the sake of having all 
computations based on an uniform set of numerical constants. 

Tablb 1. 


























































































































































' Tisserand, M^. G^l., n, p. 235. 

* Fisher, loc. cit., p. 4i7; also Kelvin and Tait, Natural Philosophy, n, p. 415. 



In table 1, fifth column, /(x) stands for (1— ix*)^ tabulated for com- 
parison because of its occurrence in an approximate formula for the density 
used later. The maximum value of g occurs at x« 0.8411, or about 630 
miles below the surface. 

Tablb 2. 






























































































































64 J2 


















































The second and fourth columns give the values of the pressure and bulk- 
modulus in millions of atmospheres, an atmosphere being taken as one 
megadyne per sq. cm., while the seventh column gives the specific com- 
pressional energy in billions of ergs per gram. The last column is added 
for the sake of vividness of interpretation in terms of temperature, being 
the measure of the thermal equivalent of e in centigrade degrees, under 
specific heat one-fifth that of water. 

Of the total store of potential energy ^, it appears that 72 per cent is 
accounted for by ^j^, which would be the energy exhausted if a system of 
planetesimals, originally infinitely dispersed, united to form a homogeneous 
sphere of density p^; this portion is to be thought of as energy of impact. 
Of the remaining portion ^— ^j^, the compressional energy E accounts for 
53 per cent. 

It remains to be determined whether the excess of ^ - ^j^ over E is also 
to be ascribed to energy of impact. That a part at least must be so con- 
sidered follows from the assumed character of the actual aggregation. For 
when a given particle impinges on the momentary "surface, the mass already 
gathered has condensed, under its own gravity, to smaller dimensions than 
it would occupy in the homogeneous sphere; the radius being smaller, the 
velocity of impact is greater than would be the case under the attraction of 
the same mass in homogeneous form with the minimum density p^. For a 
particle having any assigned position in the completed planet the attracting 
mass within can be computed by (26). But the velocity of impact depends 
also on the momentary radius of that mass, before it has been further con- 


densed by deposition of the remaining layers, so that in order to determine 
the total amount and distribution of the energy of impact it is necessary to 
know how the compression advances in the existing nucleus as material is 
deposited at its surface. 


It is possible, under the foregoing assumptions, to trace such a history 
of the accretion, if it be supposed that the equilibrium of the mass is main- 
tained at each stage of the process. For if the preceding equations can be 
accepted as approximate representation of the action of the earth-substance 
under compression, to a similar approximation equation (20) must be viewed 
as representing a physical property of that substance, independent of the 
dimensions of the mass into which it is aggregated, the constants A, q being 
physical constants of the material. The distribution of density in a body of 
any size is then represented by equation (25), without change in the value 
of q, but with the maximum value of j9 chosen to agree with the total radius, 
and with central density such as to give the surface density the fixed value 
p^. The distance of any particle from the center is then fixed by the angle 
fi, which is the measure of that distance in terms of a unit of 1/g centi- 
meters, or about 1,600 miles. 

Let r/ be the radius of the nucleus already formed at a certain epoch, 
and r^ its radius when compressed under the total load afterwards depos- 
ited; let /, r be the central distances of any interior particle at the same 
epochs; then r/ is to be determined as a function of r^, and r' as a function 
of r and r,; the latter function describing the history of the condensation 
in the sense that it fixes the position of any assigned particle at any epoch, 
if the epoch is specified by the final location of the particles which then lay 
at the surface. A translation into time-relations could then be made for any 
postulated law of variation in the rate of accretion. Let p^' be the central 
density at the earlier epoch. For brevity put 

JB=^ C^ainP-pcosP (36) 

to be indexed, like the various values of j9, by analogy with the values of r 
to which they refer, as in equation (26) . 

By equation (26) the identity of the masses within the radii r/, r, at the 
respective epochs gives the condition 

p,'C/^p,C. (37) 

similarly for the corresponding radii r', r; 

PoC'^p,C (38) 

and from the constancy of density at the momentary surface 

P,'B:^P,B, (39) 

These equations are to determine the values of p^'j /?/, ^y for any assigned 
values of /?, and /?, for instance first ^9/ from (37) and (39) which give 


then p^' from (37) and finally ^ from (38). The values of j3/ and p^' 
as functions of j9^ are given in table 3 below. The former were obtained 
by interpolation from an extended table giving C/JB and C/B| as functions 
of j9. To trace the positions of interior particles at different distances and 
different epochs would require a double-entry table, which could be supplied 
by equation (38) after p^' has been computed. 

The specific energy of impact, or kinetic energy of a unit-mass falling 
from infinity to the surface of a nucleus of mass m, and radius r/ is, then, 

e.-^ (41) 

where by (26) the attracting mass is 

wi,=— 3-0, 

The values of e^ are given in table 4, in terms of j9., which determines the 
ultimate position of the particle. 

The total energy transformed by impact is, then, 

^«-4;r / \,.^^.r*dT, 

which, since p,^p^B^ is equivalent to 

But by equation (40) 



Use of this relation to transform the variable of integration from j9, to 
P,' gives 

^' ^ j[ Wb7\P7W'^^' r' 

where the integrand is expressed entirely in terms of j9/, the upper limit 
of^the integral being unchanged according to (40). The result of the inte- 
gration is 

len^kp.'B,' r(l-ff,coti90» l-i8,coti8^ Pr\ 

which may be written: 

^^,^»^ I Wi._ 6 sinff^,cos A +|eos» ^-1} (43) 


Gomparison with (34) and (35) gives the fundamental relation 

E+Et"* (44) 

showing that under the conditions here assumed the original store of poten- 
tial energy is entirely accounted for as transformed by impact and static 
compression. The proof refers to a primitive condition of infinite disper- 
sion, but a similar conclusion would hold for any initial configuration and 
distribution of velocities provided ^ stands for the entire primitive store of 
energy, potential and kinetiC; a variation in which would make an equal 
change in £^ but leave E unaltered. 

As represented by the preceding equations the character of the process 
of accretion, upon a nucleus composed of material of definite compressibility, 
implies that the deposition of a new layer of given thickness brings about a 
certain increase of compression of the nucleus and a corresponding sinking 
of the former surface toward the center; only part of the thickness of the 
new stratum is thus effective in producing actual increase of geometric 
dimensions. In order to specify this differential depression numerically, let 
a factor of depression D be defined as the ratio, to the total thickness of 
stratum, of the part which sinks below the level of the former surface; 
then 1 — D is the ratio of geometric increment of radius to total thickness of 

For simplicity of notation let p now stand for the angle equivalent of 
the radius of the nucleus at a given epoch, the mass being by (25) and (26) : 



where p^^ is the fixed density of surface rock, so that 


dm being the increment of mass and dp the increment of radius in the angu- 
lar units defined. If, however, a stratum of the same mass dm were laid 
down without producing compression of the nucleus beneath, the relation 
to the total thickness dp of the stratum would be given by 

From these follows for the depression factor 

dp ^ ^ /pBinpy (45) 

where C is defined as in (36). The factor D is tabulated in column 6 of 
table 3, in terms of /?/, which there represents the momentary radius of the 
free surface. 


A similar situation occurs with respect to the effect of accretion on the 
moment of inertia and period of rotation of the planet. The moment of 
inertia is 

-g-TT / pr^dr 

or, measured for convenience in terms of a unit equal to -5^* G. G. S. 
units, pP 

which reduces to 

2/J*-J(6-^ (46) 

Now, in the same units the stratum itself adds to the moment of inertia 
the quantity 


.^a)9-/9«|l+(^)'|d^ (47) 

But this is in excess of the true increment, because the differential conden* 
sation diminishes the moment of inertia of the underlying mass. The true 
increment by (46) is 


which may be written: 

dl -W-2i9»K (^- 1) d^ (48) 


giving for comparison: 

which ia tabulated in column 7 of table 3. Then by combination of equa- 
tions (45) to (49) may be computed 

rflog/ j8» 1 dl .„v 

which gives the ratio of the percentage increase of the moment of inertia to 
the percentage increase of the radius. This would indicate also the percent- 
age change in the length of the day if the stratum were deposited entirely 
under normal incidence, or if the moments of momentum of the planetes- 
imals with respect to the existing axis of rotation exactly compensated 
each other. If, however, it be supposed that variations in the rotation, or 
even the existence of the rotation, were brought about by lack of such com- 
pensation, then the equations allow this effect to be distinguished from that 
due merely to changes in the moment of inertia. 


As illustration may be considered the case of superficial strata deposited 
when the earth had attained practically its final dimensions, and corre- 
sponding therefore to the last entry in the tables. A stratum, for instance, 
exerting the same pressure as a mercury column of 760 mm. would have a 
thickness of 3.8 meters; the depression factor D is 0.737; hence that stra- 
tum would depress the former surface 2.8 meters, giving actual increase to 
the radius of only 1 meter. This increase bears to the total radius a ratio 

^-—X 10~*, while the value of ., ^ is 7.08; hence, for the case of normal 

impact, the corresponding increase of the period of rotation is about 

-Q X 10~* of the whole, or 0.096 of a second for a day of the present length. 

To increase the period just one second out of 86,164 would require a 
stratum 39.7 meters thick. The illustration also indicates that under the 
assumed law of compressibility the mean pressure of the atmosphere is 
responsible for a diminution of the earth's radius amounting to 2.8 meters. 

The preceding computation of pressures and densities has been based 
on the condition of hydrostatic equilibrium, which for any substance not 
completely fluid can be considered strictly applicable only in case each por- 
tion of the mass is subjected to compression in such a way as to avoid any 
<listortion of shape, which would call into play reactions against shearing 
stresses. It will therefore be instructive to determine what kind of defor- 
mation in the elementary portions of the mass is implied in the foregoing 
account. In view of the radial symmetry assumed, the distortion at any 
point in the completed planet may be expressed in terms of a distortion- 
factor S, defined as the ultimate ratio of vertical to horizontal dimensions 
of a mass which when first deposited at the surface was cubical. 

The ratio of final to initial horizontal dimensions is PJP/, being the 
same as the ratio of radii of two spheres passing through the same particles 
at the respective epochs. The vertical or radial ratio in the same sense is 

if dp', dp represent the respective thicknesses. These give for the distor- 


^^3-^, since conservation of the mass of a stratum implies 


whose value at various depths is given in column 5 of table 4, which shows 
that the vertical compression throughout exceeds the horizontal, the differ- 
ence being most marked about one-fourth of the way to the center, where 
the ratio is 0.824. This would be a violent deformation for a body with 
perceptible rigidity, but may be admitted in the present theory if it be 
supposed that under extremely slow changes the substance is practically 
plastic, even if highly viscous. The chief uncertainty would then relate 
to the energy of compression, which should take account of the work done 
against viscosity under shear. Further comment on this matter will be 
reserved for another place. 



Tablb 3. 
























































































































































































Table 3, taking as argument the ultimate distances from the center of 
what were the surface particles at the respective epochs, shows the varia- 
tion of the relative radius, central density, mass, and surface gravity, of the 
growing planet, together with the depression factor for differential accretion, 
and the factors used in determining the effect of differential accretion on the 
moment of inertia; in every case in terms of j9/, which represents the actual 
momentary radius. 

Tablb 4. 






















408 J2 

































Table 4 is a continuation of table 3, giving in terms of the same argu- 
ment the specific energy of impact for a particle falling from infinity, in 
billions of ergs per gram, the same in centigrade degrees imder specific heat 
0.2, the surface parabolic velocity in miles per second, and the distortion 
factor indicating the permanent deformation of elements of the mass at the 
various points along the radius. 


It has been supposed in the foregoing that the distribution of heat was 
not sensibly affected by conduction during the relatively short epoch of 
accretion. If, now, it be supposed that the subsequent changes in distri« 
bution are determined by conduction only, in accordance with Fourier's 
laws, then the form of the temperature curve d^d(r,t) at each instant 
may be determined by the differential equation 

provided the form of the curve at the initial instant, say <»0, appropriate 
conditions relating to the surface, and the values of the conductivity X and 
specific heat a be assigned. Since the variations of X and a under changes 
in the physical condition of the substance are almost purely matters of 
conjecture, the chief value of such an inquiry might well be considered to 
lie in the determination of features of the thermal phenomenon which seem 
to persist imder varied assumptions on these points. Since, however, the 
method of superposition of special solutions is practically the only known 
way of obtaining general solutions of equations like (52), it wiU be supposed 
that the latter is linear, X and a being assigned in each special case as func- 
tions of r but independent of 0, and that the surface equations are linear 
and homogeneous in and its derivatives. 

According to Fourier's method, of expansion into an infinite series each 
term of which is a solution of (52) and satisfies the surface condition, the 
solution may then be sought in the form 

<?(r,0-T^»«"^»t/.(x) (^3) 



r,^^-fll (64) 

and y^ {x) or y (jx^ x) f or n = 1 . . . oo are the appropriate fundamental 
functions, which are to be determined from 

where ^(x)=^/^, <p(x)='oplo^Q, and the successive values of /x employed 
are those which allow the individual terms of (53) to satisfy the surface 
condition. The coeflScients in (53) will be given by 

r a:' f (x) y„{xy dx 



if the primitive temperature curve is given in the form 

e(rfi)^d,F(x) (57) 

0^ being the initial temperature at the center. 

To determine the form of this initial curve it is necessary to know in 
what way the storage of the compressional and impact energy is mani- 
fested in rise of temperature, which is again a matter of hypothesis. Fish- 
er's third assumption, as stated above, implies that the energy stored in 
unit-mass stands in a definite ratio to the increment of temperature. That 
ratio is, then, of the nature of a specific heat, expressed in mechanical imits, 
but is subject to question as to its identification, as is tacitly done by Fisher, 
with the specific heat in the ordinary sense, as relating to rise of tempera- 
ture due to heat transferred by conduction or radiation. It will be shown 
later, however, in connection with a detailed criticism, that the theory thus 
developed may fairly be considered self-consistent, in the light of thermo- 
dynamic laws, when associated with what appears as a certain extreme view 
regarding the thermodynamic properties of the earth-substance. An effort 
will then be made to develop an opposite extreme view, to permit comparison. 

For the present, then, the primitive temperature wiU be supposed de- 
fined by 

^-^' (58) 

where v is the fractional value of that part of the energy of impact which 
remains after the loss by dissipation at the surface, and a is the specific heat 
identical with that occurring in equation (52). A variation of a with the 
density need not be excluded, but any possible variation with the tempera* 
ture will be disregarded, not only because of the practical value of keeping 
the linearity of equation (52), but also becaiise any imcertainty from this 
source would be bound up with the inevitable obscurity of the very notion 
of temperature under circumstances so far beyond the range of laboratory 

The multiplier y may also vary for different portions of the mass, its 
value depending on the rapidity of the accretion. It could be unity as one 
extreme, for deposits made with such rapidity that each stratum is covered 
up before its loss of heat by radiation is sensible; or zero as the other ex- 
treme, for accretion so slow that the cooling of the momentary surface- 
stratum by radiation is practically complete. 

One curious possibility may be noted in passing. With sufficient veloc- 
ities of impact and appropriate values of y, it would be possible for the 
expression e+y«< to have the same value for all strata; if then a were like- 
wise constant, this would indicate a primitive temperature uniform through- 
out the body. For example, if Ci should be for each stratum that derived 
from impact at the corresponding parabolic velocity, as listed in table 4, 
and y should range from 0.38 for the central portions to 0.27 for the surface 
layer, then e+yc< would have everywhere the value 1.7X10" ergs per gram 
mass, and for a =0.2 this would give a primitive temperature of about 
20,000°. Since this value of a may be too low, and it is practically certain 


that not all of the energy can be considered as manifested in the raising 
of the temperature, this number is probably somewhat too high. But in 
any case it appears that the planetesimal hypothesis could thus plausibly 
assign an origin for precisely the kind of initial thermal condition of the 
mass postulated in Kelvin's famous theory, which would then serve to 
indicate the subsequent thermal process, and in particular the probable 
age of the earth reckoned from the close of the epoch of formation. 

It seems, however, more in accord with the spirit of the general hypoth- 
esis to assign no such relative importance to the energy of impact. For 
the value of e^ should probably be assumed much less than its paraboUe 
value, being due to the relative velocities simply of bodies which might 
partake more or less of a common motion; for instance of bodies moving 
in similar directions around the sun in intersecting orbits of moderate eccen- 
tricity. Moreover, even with what might be considered a rapid rate of 
accretion, when compared with astronomical processes in general, by far the 
greater part of the heat so generated must be expected to escape quickly 
by radiation, thus making v a small fraction. The computations following 
refer to the case when the value of ve^ is insignificant, so that the compres- 
sional energy alone is effective in producing the primitive temperatures, 
which are then to be computed from column 6 of table 2 and the assumed 
values of a. 

The same relative freedom of dissipation into space would tend to keep 
the surface at a low and equable temperature after the accretion had 
ceased. It will, therefore, be assumed that the surface temperature keeps 
a fixed value, which may be taken as the zero of reckoning, and whose 
place in an absolute scale would depend largely on the thermal influence 
of the atmosphere. It is further supposed that the rise of temperature 
produced by compression can be reckoned from this point. With these 
stipulations the solution will be determined if definite hypothesis is made 
regarding the conductivity and specific heat. 

If X and a are uniform throughout the mass, then ^(x) = 1, and ^(z) 
=^plPo- Since great precision is needless in these computations, the ratio 
plPf^will be replaced by the convenient expression (1— ix*)*, whose close 
accordance with it appears from columns 4 and 5 of table 1, and which 
simplifies the determination of the solutions of equation (55), which then 
reduces to 

the required solution of which may be written: 

2/(/i,x)=2*(-iya,x^* (60) 


where the a's, which are functions of /i, are all positive and determined by 
the simple recursion-formula 



The values of fi required for the expansion (53), with the surface con- 
dition named, are the roots, infinite in number, of the equation 

y(/i,l)-*ir(-l)'a,=0 (62) 

These roots may be determined as far as required by a trial and error tab- 
ulation of y(ji, 1), the computation yielding also the coefficients a^ but 
increasing rapidly in length with each succeeding root. The first two roots 
and the corresponding fundamental functions are: 

/ij- 13.0689 

y,(a;)-l -2.17648 x»+2.07407x*- 1.39934 a;*+.72867a:»-.31419x*» 
-f. 11658 x«-.03812x" + .01120x**-.00300x^" 
+ .00074 x»-.00017x» + .00004x"- .00001 x*« .... 

/c, -55.313 

y,(x) - 1 - 9.21889 x» + 28.26204 x*- 49.69103 x* + 61 .65731 «• 

-59.54423 x*«H-47.37952 x"-32.22346 x" + 19.21509 x** 
- 10.23514 x«H- 4.93949 x»- 2. 18394 x» + . 89259 x" 
-.33971 x» + .121 14 x»-. 04068 x« + . 01292 x» 
- .00390 x»* + .001 12 x«»- .00031 x» + .00008 x« 
-.00002x^ .... 

The number of terms to be included and the magnitude of the individual 
coefficients increase rapidly with the index of the component, so that if 
many components are of sensible influence in the representation of the 
primitive temperature curve, the accurate determination by this method of 
the transformations produced by conduction would require computations 
of serious length. But the effect of the higher components on the general 
features of the cooling process can be conjectured, with the aid of certain 
general properties of the fundamental functions which are obvious in the 
light of the theory of linear differential equations of the second order. 

The constants fiv p^t * - * will show a rough proportionality to the 
squares of the natural numbers. The solutions y{pL^ x) or yjjt) will be 
oscillatory in such wise that y^ has n — l roots between and 1, the inter- 
mediate half-waves having amplitudes all less than the central amplitude, 
which is unity, and decreasing in order when counted from x=0 toward 
x« 1, exhibiting these features in a more marked fashion than the functions 

, because of the fact that the coefficient tplx) decreases as x goes 

from to 1. 

For constant a the primitive temperature curve is obtained from the 
curve for e/e^, which has unit amplitude at the center, by multiplying 

all ordinates by the central temperature, -j; which for <r = 0.2 is about 
20,000^, the temperatures being then given by column 8 of table 2.^ It 

* The temperatures given by Fisher are larger than those here listed in the ratio of the 
number of pounds in a cubic foot of water, presumably through some inconsistency in the 
units used m the equivalents of formulas (21) and (22), which however does not occur in 
the corresponding npressions in his "Physics of the Earth's Crust," p. 29. 


will be convenient however to express the temperatures and coefficients 
A^ in terms of the central temperature as unit, to leave free choice as to 
the numerical value of a. The coefficients A^, A^ then have the values 
named below, computed according to (56) by mechanical quadrature. 

The influence of the first two components appears from table 5. The 
second and third columns give the first two fimdamental functions, the 
fourth the primitive temperature curve, and the next two the terms out- 
standing after subtraction of the first and of the first two components re- 

Taslm 5. 


















— .0523 

— .0362 

— 9.97 





— .0480 

— .0330 

— 9.89 





— .0413 

— .0280 

— 9.76 





— .0326 

— .0214 

— 9.67 





— .0225 

— .0139 

— 9.33 





— .0120 

— .0050 

— 9X» 





— .0019 

+ .0017 

— 8.67 





+ .0070 

-f .0083 

— 8J24 



— .0343 


-f .0135 

-f .0129 

— 7.74 



— .1222 


+ .0175 

+ .0155 

— 7.16 



— .1826 


+ .0186 

+ .0156 

— 6.60 



— .2164 


-f .0160 

-f i)124 

— 6.78 



— .2266 


-f .0102 

+ .0065 

— 4.82 



— .2176 


+ .0005 

— .0031 

— 8.76 



— .1937 


— .0092 

— .0124 

— 2.47 



— .1599 


— .0199 

— .0225 

— 0.86 



— .1205 


— .0291 

— .0311 

+ 1J21 



— .0790 


— .0328 

— .0341 

-h 4i)l 



— .0382 


— .0260 

— .0266 

+ sm 









F (x) = W„ the primitive temperature curve; 
F,(x) "F (x)- Ajy,(x) A^ =1.0537 

Ft(x) =Fj(x) -il,y,(x) A, = - 0.0165 

In the seventh column -rr is the initial rate of change of temperature 
in convenient units according to equation (65) below; these entries, to be 

reduced to absolute units, must be multiplied by 



as they stand the 

umt of change would be about 3i^ in a billion years, with the numerical 
constants used. 

The results show that the first component is by far the most important; 
its amplitude at the center differs from that of the temperature curve by 
only 5i per cent, and the divergences fall below that percentage over a 
range of more than eight-tenths of the radius from the center. If this com- 
ponent alone were present the temperature at any time would be represented 


exhibiting the aimple type of cooling described by Fourier, where conduc- 
tion produces simply a progressive diminution in the ordinate scale of the 
temperature curve, according to the time-exponential law, without change 
in the ratios of the ordinates. Since y^ix) is everywhere positive, this 
would mean an actual decline of temperature at each point proportional 
to the temperature itself. The gradient at the surface would be initially 
1^ in 417.5 meters, for <7""0.2, and would decline according to the same 
law as the temperatures themselves. This shows that under the condition 
for the moment assumed only a small part of the present observed gradient, 
about 1^ in 30 meters, could be ascribed to this component, unless the 
specific heat were taken very low. 

The time-rate of the process is specified conveniently by means of the 
interval r, which is the time required to reduce the amplitude of the corre- 
sponding component to - or 0.368 of its primitive value, and is to be deter- 

^ k 

mined by (54), when the value of — , the "thermometries conductivity, 


is assigned. If the latter be taken in the neighborhood of 0.01 for surface 
rock, the value of -~^ being one-fourth of this, the value of r^ is about 

4 X 10" years. The time required to reduce the amplitude of the first com- 
ponent by 1 per cent would be about four billion years. 

Any higher component dies out in a similar way, at a rate indicated by 
its value of r; but because of the alternation in sign of the fundamental 
function, would, if occurring singly, indicate falling and rising of tempera- 
ture for successive zones in alternate order along the radius, the number 
of zones being equal to the index of the component, with the central tem- 
perature falling or rising according to the positive or negative sign of the 
coefficient A. Thus the second component has a negative coefficient, in 
magnitude less than one-sixtieth that of the first, but with r, somewhat less 
than iv^; since y, is everywhere numerically less than y^, this means that 
with respect to changes of temperature the second component simply modi- 
fies the e£Fect of the first nowhere to an extent more than one-fifteenth of 
the total e£Fect due to the latter alone. In the zone extending 0.43 of the 
radius from the center the temperature falls somewhat more slowly and 
thence outwards more rapidly with only these two components included 
than would be the case with the first alone. 

The influence of each further component could be traced in a similar 
way, and many would doubtless be foimd to be sensible within the range 
of accuracy of the tables above, if the computation to that degree of ac- 
curacy should prove to be feasible. But in the absence of simple analytic 
expressions for the functions involved it would be necessary to do this by 
numerical calculations of extreme length on accoimt of the greater and 
greater number of the coefficients a^ needed, and the insufficiency for deter- 
mining the coefficients A^ of a tabulation of the fimctions with a moderate 
number of entries. 

The residuals in column 6 show that the influence of the higher compo- 
nents is meager in the central portions, but relatively serious in the more 


superficial zones. It is accordingly in the latter region that such general 
considerations as above are insufficient to give a just idea of the complete 
thermal process. 

Conjecture may, however, be made with some confidence as to the prob- 
able character of the modifications produced by the higher components, by 
direct inspection of the primitive temperature curve, which by (22) and 
(58), with the substitution of the approximate expression used previously 
for the density, becomes 

e =i?o(i -^T(i - \xy{i~x')-^ (64) 

which is accurate enough for the purpose. Then, according to (52), the 
initial rate of change of temperature becomes: 

bO ex 

dt a^jr 



/9 X 
in which the coefficient of — ^-S is that tabulated in column 7 of table 5. 

For the part of-rr due to any single component, say the nth, the corre- 

'^ at 

spending coefficient would be fi^A^^. 

These figures indicate that from the center outwards over a distance of 
about eight-tenths of the radius the thermal process is not very different 
from that represented by the first component alone, except that the latter 
exaggerates the rate of decline somewhat in the more central portions on 
account of the opposing effect of the higher components. But in the outer 
zone of about two-tenths of the radius the process is in the earlier stages 
totally different; here the temperature actually rises for a certain interval 
of time, which would be different for different depths, very short for points 
extremely near the surface because of the constancy of the surf ace-tempera* 
ture, and for points near the boundary between the regions of rising and of 
falling temperature, but presumably of considerable length at intermediate 
depths. Since this trend of temperature in the outer zone is brought about 
by the higher components, which practically die out in a time sufficient to 
produce only a relatively small change in the amplitude of the first com- 
ponent, the whole process may be conceived to occur in two epochs, an 
earlier one of gradual accommodation of the temperature-curve to the slowly 
declining first component, and a later one where that component is left 
practically isolated. During the latter epoch the temperature would decline 
steadily at all points at rates nearly proportional to the existing tempera- 
tures. But during a large part of the earlier epoch the heat lost from the 
central portions is conducted through an intermediate zone, whose thermal 
condition is nearly stationary, and thence outwards to produce an exaltation 
of temperature in a zone a few hundred miles thick just below the surface. 
Dissipation through the surface in the earlier stages is very slight, owing to 
the smallness of the gradient, the primitive temperature-curve being tan- 
gent to the X-axis at the point a;= 1, because of the occurrence of the squared 
factor (p—py in formula (22). 


The zone of rising temperature, which would be narrowed down and 
finally disappear as the conduction progressed, extends at the start from 
the surface to a depth of about 700 miles. The total rise of temperature 
would be trivial near the extremes of this zone, but more marked toward 
its interior. According to the residuals in column 6, table 5, the rise may 
be expected to be most significant in the neighborhood of a depth of about 
400 miles, reaching in that region, with allowance for even a considerable 
percentage of decline in the first component, a value of probably at least two 
one-hundredths of the central temperature, or 400^, which is the increase 
over initial temperatures ranging about 1,200^. Such a change might 
carry the substance through its temperature of fusion, even under the high 
pressures there sustained. 

The foregoing sketch of the thermal process lacks, of course, the pre- 
cision which could be reached through a computation extended to include 
all components of sensible influence; this also would alone suffice to yield 
an accurate estimate of the time-intervals implied, which would probably 
be counted in billions of years for the epoch during which rising tempera- 
tures occur. It must be noted also that a small outstanding portion of the 
energy of impact might alter the features of the thermal process seriously, 
especially in the strata near the surface, where the very fact of the oc- 
currence of the rising temperatures may be said to be due to that relative 
deficiency in the heat from the purely compressional source which is rep- 
resented by the upward concavity of the initial temperature-curve. 

It is desirable, however, to know more precisely the result of the complete 
computation, under at least one set of reasonable assumptions, which, in 
the absence of experimental information, may fairly be conditioned by the 
practicability of the calculations; for instance, through the use of a suitable 
alternative hypothesis regarding the specific heat. This has thus far been 
treated as a constant, but there would seem to be some reason, under a 
molecular theory, for supposing it to decrease with increase of density, since 
the consequent diminution of the intermolecular spaces might tend to throw 
more of the energy into the " unordered " kinetic or thermal form, by inter- 
fering with the " ordered " movements which have been conceived by Hertz 
and others to accoimt for the storage of energy apparently in latent or 
potential form. It will be of interest to see how far the thermal process 
described above is modified by supposing the specific heat thus variable. 

A simple supposition on this point, hardly more arbitrary than any other 
that could be made and having at least the merit of yielding tractable form- 
ulas, is that a is inversely proportional to /o, or ap^a^^. With constant 
" calorimetric " conductivity k this makes the "thermometric" conductivity 

— also constant, and reduces equation (55) to the form: 

giving for the fundamental fimctions: 

smnnx (87^ 

^ nnx 


and for the expansion (53) the ooefficienta: 


r I zF(z) ai 

il»->2n;r I zF{z)amnKxdx (68) 

if the initial temperature-curve is y^dj^ix) and the coefficients are ex- 
pressed in terms of 0^ as unit. The intervals r are determined by 

'-'-^ w 

and are thus in this case strictly proportional to the squared reciprocals of 
the natural numbers. 

To determine F{x), it may be noted that the initial temperatures can be 
determined from those of the previous hypothesis by multiplying by pipi] 
giving for the central temperature 

which for a « 0.2 is about 81,000^; and for the curve in terms of the central 

i-n-^-i^) (7.) 

For convenience in determining the coefficients A^ the last equation will 
be replaced by the formula 

F(a:) =(1 -a:»)»(l-y x«+|x*) 

F{x) =l-^x«+||x*-gx*+|a:» (72) 

which, in comparison with (71), leaves residuals at most 3 units in the 
fourth decimal place. By (68) the expansion has then the coefficients 



Sn^'^ — I f ^ sin n^d^ 

which can be reduced through integration by parts, giving finally 

A Of nn-iJ 136 562 6480 . 72576 ) .„. 

The first few of these coefficients are: 
^1= +0.70590 i42= +0.36797 4^= +0.05354 ilo= +0.02240 
^4,= -0.10765 i45= -0.03284 A^^ -0.01632 



The alternation of the sign begins with the second and the steady decline 
in numerical value continues throughout. 

A convenient unit of time for exhibiting the changes produced by con- 
duction in the temperature-curve is the interval T required to reduce the 
amplitude of the first component by 1 per cent, which with the same con- 
stants as before, for the surface-stratum, would be about 1^ billion years. 
The successive changes in the temperature-curve are shown in table 6, 
abbreviated from an extended computation covering an epoch of 35 such 
intervals, the number of terms of the series included ranging from 100 for 
the earliest entries to 5 for the latest. The unit of temperature is the initial 
temperature at the center. 

Tabub 6. 



• (6D 

• (lOT) 

• (lAT) 

• (2or) 

• (262*) 

• (SOD 

• (867) 






















































































































































































Here, as in the previous case, the earlier stages of the conduction are 
marked by the division of the mass into two zones, an inner where the 
temperature falls and an outer where it rises. The spherical surface of 
boimdary between these lies initially at a depth of not quite 1,200 miles, 
and as the conduction progresses rises toward the surface. Its passage 
through any particular horizon marks the attwiment of the maximum tem- 
perature which occurs at that depth; while its ultimate coalescence with 
the free surface, simultaneous with the occurrence of the maximum surface 
gradient, indicates the final establishment of a downward trend of temper- 
ature throughout the body. This decline during the further progress of the 
conduction is in the outer portions first accelerated and later retarded, 
and is ultimately everywhere of a character more and more closely approx- 
imating to the simple type represented by the first component alone. 

Table 7 gives numerical data, with the epochs reckoned from the begin- 
ning of the conduction: 


Tabui 7. 


in miles. 

Epoch of 




Total rin. 



16.9 T 
22.2 T 
25.6 7 
27 AT 
29.1 T 







The zone of rising temperatures vanishes at epoch 29.7 T, correspond- 
ing to the maximum surface gradient of 1^ in about 180 meters, about one- 
sixth of that obtained from observations at the present time. The great- 
est total rise of temperature occurs at depth about 480 miles, where it is 
2,200^, with initial temperature 2,340°. 

As compared with the former case sketched above, the results of the 
present assumptions, though giving considerably larger primitive temper- 
atures, show a close qualitative similarity in the thermal changes. But the 
zone of rising temperatures is at first somewhat deeper, the temperature- 
increments within it are greater, and the period of its existence longer. This 
is to be ascribed largely to the influence of the second component of the 
series, whose coefficient is here large, while in the former case it is relatively 
trivial. In both cases the conductivity has been treated as constant. It has 
often been supposed, however, that it would probably increase considerably 
with increase of density. The effect of this would be to facilitate the trans- 
fer of heat from the central parts to the outer zone, presumably increasing 
the total temperature-increments there, perhaps shortening the time dur- 
ing which the temperature rises. It would be likely, also, to increase the 
value of the maximum surface-gradient, but whether this could be brought 
up to the present observed value through any reasonable assumption of 
this character remains to be determined. 

The magnitude of the thermal changes in the zone of rising tempera- 
tures, resulting from the arbitrary special conditions developed above, seems 
to make it probable that the existence of this zone should be considered an 
essential feature of the thermal process under the planetesimal hypothesis. 


In the development of the theory in Part I, the attempt was made to 
give explicit statement, though without critical setting, of the more impor- 
tant of the secondary hypotheses which were used as auxiliary to the gen- 
eral hypothesis in order to give the theory a sufficiently definite form. But 
since the main purpose is to seek such features of the geophysical phenom- 
ena as seem to be essentially consequences of the general hypothesis, it is 
necessary further to inquire how far the results are peculiar to the special 
conditions adopted and how far they seem to persist under variations of 
these secondary assumptions; and also to what extent these assumptions 
are subject to obscurity or positive objection through the accessibility of a 
direct or indirect test by observation or well-established theory. 

The Laplacian law of density has been assumed chiefly because of its 
analytic convenience, though it seems doubtful whether any geophysical 
theory is likely to be sufficiently trustworthy in detail to afford more than 
a crude test of any assumption on this point. Nevertheless, by inspection 
of some of its consequences, it is possible to surmise the probable character 
of its departure from the true law of compressibility of the average earth- 
substance. It appears that the modifications which seem to be needed 
from the standpoint of the planetesimal hypothesis agree, at least in kind, 
with those familiar from the indications of general geophysical theory. 

Colunm 7 of table 3 shows that in the earlier stages of the growth of the 
planet the thickness of a stratum deposited is nearly all effective in enlarg- 
ing the geometric radius of the mass, while toward the last not much more 
than one-quarter of a new stratum remains above the former horizon; as 
the mass grows larger the less significant becomes the actual increase of 
dimensions produced by a new stratum of given thickness. Moreover, with 
a strict interpretation of the law of compressibility assumed above, there is 
a definite limit to the possible radius of the planet, no matter how much 
material might be laid down. For in equation (25) the factor 9 is a defi- 
nite constant, depending solely on h, which is determined by the compressi- 
bility as a physical constant of the material, independent of the dimensions 
of the mass into which it may be aggregated; while if there are to be no 
meaningless negative densities introduced, the angle /9 can not surpass the 
value K, at which the density at the center becomes infinite. This means 
that no amount of accretion could produce a mass with radius greater 

than ^ centimeters, or about 5,000 miles. 

Now, even independently of any supposition as to the actual origin of 
the planets, there seems to be little reason for supposing that if deposits of 
indefinite extent could be brought about at the surface of an existing planet 
there would be such a limit of growth, at least of such comparatively meager 
dimensions. This objection has not much force, for the reason that the law 
might be practically accurate for the range of densities contemplated and 
seriously in error for the higher densities; but its suggestion is that the true 



compressibility most probably diminishes, as the density increases, more 
rapidly than is postulated by the Laplacian law. This would allow a larger 
part of the compression to take place during the earlier stages, giving to the 
nucleus, when it reaches a given radius, a larger mass than is assigned it 

In the light of the planetesimal hypothesis a similar concluaon can be 
reached in another way. For in view of the like origin predicated of the 
earth and other bodies of the solar system from the primitive planetesimals, 
composed of more or less similar materials of definite compressibility, it may 
be expected that at the successive stages when the earth-nucleus reached 
dimensions equal to those of various planets at the present time its mass 
should show some agreement with the observed masses of those planets. 
The supposition of the small upper limit of diameter just mentioned is of 
course negatived by the existence of the planets of the Jovian group. But 
such comparison is futile, partly because of the uncertainty as to their 
true dimensions, brought about by their extensive atmospheres, pardy 
because of the wide difference in physical condition as compared with the 
earth, illustrated in particular by mean densities smaller than even the 
surface-density assumed for the earth. Moreover, it is conceivable that 
the discrepancies of an assumed pressure-density law might become serious 
only outside of the range met with in smaller bodies like the earth. But a 
comparison with the other planets of the terrestrial group should prove 
instructive. Table 8 gives their observed radii and masses as compared 
with the earth, and the hypothetical masses computed by interpolation 
from columns 2 and 4 of table 3. 

Tabub 8. 













To interpret this table it is necessary to observe that if the masses cor- 
responding to assigned dimensions be computed according to two different 
laws of compressibility, using the mean and surface-densities of the present 
earth as given constants, then those masses must approach agreement, on 
the one hand for small bodies where the compression is slight and the aver- 
age density therefore not very different from the assigned surface-density, 
on the other hand for bodies approximating the earth in size, the average 
density then approaching the assigned mean density. Thus it appears that 
the divergence of result between two such assumptions would be likely to be 
most marked for bodies of intermediate size, like Mars, Mercury, and the 
moon, while the agreement should be close for Venus on the one hand and the 
asteroids on the other — as, for instance, if the masses resulting from the 
Laplacian law be compared as above with the observed masses, supposing the 
latter to correspond to a definite, though unknown, law of compressibility. 


The masses of the asteroids are purely conjectural, while that of Mer- 
cury is too uncertain to be used. For Venus the computed mass exceeds 
the observed; but at least part of the difference may be ascribed to the 
influence of atmosphere and irradiation in bringing about an overestimate 
of the planet's true diameter; for example, if the observed radius were 
corrected by — 2 per cent of its value, the computed and observed masses 
would agree. For the moon and Mars, the computed masses fall below 
the observed by 15 per cent and 20 per cent, respectively. 

From this, as before, the suggestion is that the assumed law of com- 
pression should be modified in the direction of allowing greater compressi- 
bility at the lower densities, and less at the higher, which would have the 
effect of assigning greater mass to the nucleus at intermediate sizes, and to 
the present earth a steeper density-gradient near the surface, together with 
a relatively more nearly homogeneous central portion. This agrees with 
conclusions which have been drawn from observations of precession and the 
transmission of seismic disturbances. 

In view of this comparison it seems quite conceivable that a law of com- 
pressibility might be constructed, agreeing with the data furnished by the 
earth in its present condition, and such that the observed masses of the 
planets now existing in its neighborhood would prove to be the same as 
those computed for the nucleus at epochs when the din^ensions correspond. 
It wiU accordingly be of interest to review the previous theory, with the 
substitution of a density-formula whose variations from that of Laplace 
have the general trend indicated. 

A formula of this character for the density is the simple one proposed 

by Roche: ^ ,- .v . ,^-v 

jO— ft(l— car) c ""Const. (74) 

From this, according to the general equations previously used, are derived: 

4 / 3 \ 

m - g^ ;r/>o r» ^1 - y cx^J (75) 

whence H is determined by (15); also: 

p.d|(«»_«^») + (a._«j.)l (77) 

A , .. a+2«,+2 ,_„ 

'-^(*-^«) •— T— ^78) 

/o«-ft(l-|c) p,~p,(l-e) (79) 

E=8itAr*'^e(l~e) (80) 


87rAr,«(l-|c+jC») (81) 

* Roche, Acadteie dee Sdenoes et Lettres de Montpellier, v. 8, 1848, p. 235. 


in which are put 

^ "16 ^^P%W 



The constants p^j c may be determined by assigned values of the mean 
and surface densities, which will here be supposed to be in the ratio 2:1, 
giving c = ^, whence results 

/Oo = 9.653 

/t)i- 2.758 


Hi -3.45X10" 
*- 2.426X10" 

and for the quantities p, p, H at various depths the values listed in table 9, 
which for the present density-formula replaces the corresponding columns 
in tables 1 and 2. This value of c is somewhat less than that used by Roche, 
which seems to give rather too small an estimate for the surface-density, 
when the more modem determinations of p^ are used. 

Comparison with the tables of Part I shows that the range of values in 
density, pressure, and specific compressional energy is in each case some- 
what less than under the previous conditions, but that the modulus of com- 
pression, while somewhat less at the lower densities, is decidedly greater at 
the higher, showing that the departures of formula (74) from (25), which it 
replaces, have quaKtatively the character shown to be needed. Of the 
energy-totals, tfj^ is necessarily the same as before, while # and E are respec- 
tively 1.2 per cent and 14.1 per cent less than their former values, so that 
relatively more of the primitive energy is transformed by impact and lost 
by radiation. This is obviously due to the greater mass of the nucleus, at 
a given radius, and the correspondingly larger velocity of impact in thb as 
compared with the former case. 

Table 9. 















































































































Table 9, based on the formula p^Poil-j^x*), gives the values at va- 
rious depths of the density, of the pressure and bulk-modulus in millions of 
atmospheres, and the thermal measure for (7 = 0.2 of the specific compres- 
sional energy. 

To estimate the modifications in the thermal process due simply to the 
change in the formula for density, it will be well to repeat the former alter- 
native hypotheses regarding the specific heat. 

If a, X are constant, the primitive temperature-curve by (78) is 

(?-(?,(l-x»)»(l-ix»)^(l-|a:») <?,-^ (82) 

which is tabulated in column 5 of table 9, while equation (55) reduces to 

il(^l)+^<i-''^)''-o («3) 

giving for the fundamental functions 

y-2*(-l)'a,x« (84) 

where the coefficients a^ are determined successively from 

o^-l ai=|^ g<" 2t(2i+l) ^^*~^"^^^^^ ^^^ ... .00 (85) 

with the values of /i which make y vanish at a;» 1. The complete computa- 
tion to sufficient accuracy to determine the details of the conduction is 
again hardly practicable, but the first two fundamental functions with their 
parameters /i, r, and coefficients A determined from (56) are: 

yi-1 -2.04360 x« + 1.69080 a:*- .91977 x«+.36231x»-. 11362 a:**^ 
+ .02927 x« - .00645 x" + .00123 x»« - .00021 x" 
+ .00003 X**— .... 

/ii - 1 2.2616 A 1 = 1 .089 0o r^ - 400 billion years 

2/a - 1 - 8.59393 x» + 23.99826 x' - 36.99908 x« + 38.77346 x» 
- 30.56380 X** + 19.25673 x"- 10.08879 x" + 4.52008 x»« 
- 1.76799 x»»+.61344 x»- .19120 x« + .05409 x»* 
-.01400 x»*+.00334x«-.00074x» + .00015x« 
-.00003 x»*+ .... 

/£, =51.5636 A, - -.0705 6. t, =96 billion years 

These are given in table 10, columns 2 and 3, together with the residuals 
from the primitive curve due to components higher than the first and second 
respectively in columns 4 and 5, the tabulation in all cases using as unit the 
primitive central temperature 0^^, which for a =0.2 is about 16,600®. The 
last column of that table gives the initial rate of change of temperature, 
computed from 

dd eji { 312^1268 , 132 ,,,^ . 100 .) f, 5 ^^ ,^7. 

^"^r^"^"49-^--^^+'2"-i9^ri'-7^1 ^^^^ 



which follows from (52) and (82), the tabulation using, however, as unit of 


measure the value of — ^, equivalent to a change of about SJ® in a billion 

years. ^^* ^ 

Table 10. 














— 8.91 





— .0871 

— .0181 

— 8.90 





— .0817 

— .0171 

— 8A5 





— .0731 

— .0154 

— 8.76 





— .0618 

- .0130 

— 8.64 





^^ .U4o4 

— .0098 

-— 8.48 





— .0338 

— .0059 

— 8.29 





— .0190 

— .0016 

— ao6 





— .0050 

+ .0028 

— 7.77 



— .0074 


-f .0072 

+ .0067 

— 7.45 



— .1009 


-f .0170 

+ .0099 

— 7.06 



— .1679 


-f .0234 

-h .0116 

— 6.62 



— .2083 


+ .0260 

-f .0113 

— 6i)8 



— .2243 


4- .0243 

+ .0065 

— 6.45 



— .2194 


+ .0185 

+ .0030 

— 4.66 



— .1981 


-h .0091 

— .0049 

— 8.64 



— .1663 


— .0029 

— .0145 

— 2.27 



— .1255 


— .0154 

— ,0242 

— 0.26 



— .0827 


— .0246 

— .0304 

H- 8.00 



— .0401 


— .0240 

— .0268 

4- 9.01 



— .0000 





In table 10, based on the hypothesis c^c^f ^^Kf P^P$0'~"n^f ^'^ 

functions y^ y^ are the first two fundamental functions; F{x) gives the 
primitive temperature-curve, Fi(a:)=F(a;)— A^y,, F^{x)^F^{x)'-A^^; in 
each the temperatures and coefficients A are expressed in terms of the unit 
= 16,600^. The last column gives the initial rate of the change of temper- 
ature in terms of a unit corresponding to a change of 3}° in a billion years; 

the entries in this column must be multiplied by — ^—^ to be reduced to 
absolute units. ^* * 

Comparison of this table with table 5 shows that the change in distri- 
bution of the density makes no striking change in the thermal process. The 
second component has a larger coefficient, and the residuals left by the first 
two components are somewhat smaller. But these residuals have in gen- 

eral the same trend, and together with the values of -rr indicate the same 

initial partition of the mass into inner and outer regions of falling and rising 
temperature, respectively. The outer zone is, however, shallower than under 
the former case, extending at the start to a depth of not quite 600 miles. 

The increase in the tabulated values of the initial rate -rr and of the gra- 
dient of the first component for points near the surface, together with the 
coefficient A,, is offset by the decreased value of d^j in terms of which the 
entries are expressed, so that the maximum temperature-gradient at the 
surface occurring when the zone of rising temperature disappears, would be 



probably about as before. The time-scale of the phenomena can be con- 
sidered practically the same as before, especially in view of the uncertainty 
as to the appropriate values of the thermal constants. 

Under the other supposition, that X and ap are constant, the funda- 
mental functions are again defined by (67), while the primitive tempera- 
ture is 1 

5-5,(l-x«)»(l— ^x«) 

where now 


^ f5 





or about 58,000^. The coefficients of the series are given by 





the first few of which are : 

Ai« 0.80006 A, -0.29146 A^= 0.08054 A, -0.03598 

A,- —0.14122 A,- —0.05174 A,- —0.02646 

the alternation in sign and steady decline in numerical value continuing 

The time-interval T, during which the first component declines by 1 
per cent, is the same as with the former density formula, or about 1} billion 
years, with the thermal constants assumed. The effect of the conduction 
in modifying the distribution of temperature is indicated in table 11, for 
epochs differing by five such intervals. To the order of accuracy for this 
table the numbers of components of sensible influence range from about 100 
for the earliest epoch to 4 for the latest given. 

Tabud 11. 



• (5r) 

• (lor) 

• (153*) 

• (2or) 

• (253*) 

• (SOT) 

• (853*) 























































































































































































Here the zone of rising temperature extends at first to a depth of a little 
over 1,000 miles, and the chief features of its history may be thus sum- 

Table 12. 


in miles. 

Epoch of 




Total rise. 



8.3 r 
16.3 r 

20.1 T 
222 T 

23.6 r 







The rising temperatures finally cease at epoch about 24 T, with the at- 
tainment of a maximum surface gradient of 1^ in about 220 meters. 

To complete the sketch of the modified theory, in a way parallel to that 
of Part I, there might be traced an analogous history of the nucleus during 
the period of accretion, supposing the compressibility of the substance to 
correspond to equation (76). But in contrast with (25) it must be noticed 
that this could not be done by treating (74) as the general expression for 
the density at every epoch. For by (76) the modulus of compression has 

*''*'°™ H^hp^hy (91) 

and if this is to be definite and characteristic of the substance, not depend- 
ent on the dimensions of the mass into which it may be gathered, the 
coefficients A, h' must be numbers depending only on the units of measure 
used, which on account of the manner of th^ir dependence on p^ and c can 
be so only if the latter also are particular numbers in the same sense, 
not arbitrary parameters depending on the size of the body. The general 
expression for the density would have to be sought by integration of equa- 
tion (6) with the condition (91), which would yield a family of curves cor- 
responding to the nucleus at various stages, but of which only the final one 
has the simple parabolic form defined by (74). The general integral would 
contain two constants, but one may be considered eliminated either to 

make the density finite at the center or by the condition that -r- vanish 

at the center. For this particular law of compressibility the general inte- 
gral does not appear to be known in form sufficiently simple to make the 
computation practicable. 

This suggests that in connection with the planetesimal hypothesis it 
would be convenient to have a series of examples of curve-families, each of 
which could be considered to represent the distribution of density at the 
various stages of aggregation of a substance with a certain definite law of 
compressibility — equations simple enough to allow the detailed computa- 
tions, but so varied as to admit of choice in adapting to other accepted 
data. As a help towards the discovery of such it may be asked what general 
condition a family of density-curves must satisfy in order that a definite 
compressibility may be deduced. 



This condition means that if the curves be supposed given in the form 

P -nr,a) (92> 

where a is the parameter of integration, whose value corresponds, for 
instance, to the total radius or the central density, then the modulus H 
deduced therefrom must be a function of p only, independent of a. Equa- 
tions (3), (4), (5), (15) show this to be equivalent to the condition thai 


— -j- shall be a function of p only, or in terms of the functional determinant 


in which is put, as before, 

3 / m \ 


m=«4;r I pr^dr 





By differentiation and elimination of m this fundamental condition can be 
reduced to a partial differential equation for p, in rather cumbersome form. 
Trial shows that (74) is not a solution for any manner of dependence of p^ 
and c on a; but (25) is a solution, if ^^ be a function of a, and q a numerical 

Another special solution of (93) in simple form is found to be 



/I » const. 

for which 




so that for various values of a it would give the density curves for masses of 
different dimensions, if the substance were to satisfy the condition 






With such a substance the compressibility at various densities would be such 
that not only would there be a definite limit to the dimensions which the 
mass could attain, but this would be reached with a finite mass, and beyond 
that point any further addition of strata at the surface would result in an 
actual decrease in size, as appears from the following analysis. 

If p^, PifT^he the mean and surface densities and total radius at any 
chosen epoch, and a^ the corresponding value of a, then 







^ = l+c (99) 


Then since the surface density is not to change, the total radius r/ and the 
parameter a at any epoch are related by 

which, combined with (98), gives 

/icA*-(l+c)A+l=0 (101) 

a quadratic equation for il, in which are put 

Hi)' '-id' 

The condition for reality of the roots shows that the maximum possible 
radius is 

, r. 1+c 

2 • JJ (102) 

as may be deduced also from the transformed equation 

B . (1+y-l (103) 

From which it is seen that i2 as a function of A has a maximum value 

n -f c)* 2 

-R =» . at il = .. y As A increases indefinitely from its smallest 

admissible value, which is t~7~; the mass also increases from zero without 

' 1+c' 

limit, but the total radius increases to the maximum value indicated and 
then decreases toward zero. For any assigned radius less than the maximum 
there would thus be two possible distributions of density, giving masses 
less than and greater than the critical mass, and with mean densities less 
and greater than double the surface density respectively. 

If, however, the constants a^, c be determined by (97), (98), from the 
values of p^, p^ heretofore assumed for the earth, the computation ac- 
cording to (94) shows that for some distance downward from the surface 
the density would increase more slowly than with either of the formulse 
used before, but then rise more rapidly and at the center reach a value of 
15 to 16. In view of the probable corrections mentioned above as needed 
to change the Laplacian formula into one agreeing better with the data 
available, it appears that in comparison with (25) the formula (94) is much 
less satisfactory, as an approximation to the general distribution of density 


within the earth. It is here dwelt on briefly for the sake of the comparison 
it affords with still another now to be considered, which also belongs to the 
class for which , 

f =V (104) 

of which (20) and (96) are examples. 

It was shown in Part I that, with the particular law of compressibility 
there postulated, the progressive condensation under the increasing load of 
the material gathere4 at the surface would be accompanied by a deforma- 
tion of the elements of the mass whose final amount is indicated by the 
distortion-factor tabulated. Attention was called to the consequent uncer- 
tainty introduced into the determination of the work of compression if it 
be supposed that the substance offers appreciable resistance, either elastic 
or viscous, to shearing stresses, so that the working pressure is not purely 
hydrostatic. It is conceivable, however, that this deformation might be 
widely different in character and amount under another acceptable pres- 
sure-density law, so that the acceptance of the special formulas (20) and 
(25) would lead to no just estimate of the essential obscurity in the theory 
from this source. 

As a guide to conjecture on this point it will be of interest to determine 
whether there could be a law of compressibility assumed of such nature 
that the condensation of the mass would lead to no such deformation, but 
rather that the compression would be at all points purely cubical. 

If a; be an auxiliary variable, determining the location of a given par- 
ticle at some chosen epoch, for instance as before the ratio of the ultimate 
distance from the center to the total radius, then the distance r from the 
center at any epoch may be considered as a function of a and x: 

r:^(p{a,x) (106) 

Let x', x^' be the values of x corresponding to two chosen particles; then 
the mass of the spherical shell whose bounding surfaces pass through those 
particles will be ^ 

m{x^,7f)»4^n I pr^^dx (IM) 

Jx' ^^ 

As the condensation progresses the spherical surfaces will shrink, but the 
mass between them must remain constant. This means that the integral 
(106) must be independent of a, whatever the values of x', x"; this gives 
the condition 


which reduces to 

in which is put 

[dp d{pP)] 
Ida"^ dr J 

+2^P-0 (107) 

P^<^^r)^^ (108) 



obtained from (105) by differentiation and elimination of x, which is admia- 
dblei since r is a monotonic function of x. 

Moreover, the variations in horizontal and vertical dimensions of a 
given element of the mass are proportional to the variations of r and of 

-z— respectively as functions of a. The condition of no distortion demands 

1 dr 

therefore that — -r- shall be independent of a, which gives 

~--^-0 (109) 

BO that P may be written in the form 

P'^-rQia) (110) 

showing that, to satisfy the conditions named, a differential accretion at 
any epoch must depress each particle an amount proportional to its dis- 
tance from the center, but so that the factor of proportionality depends in 
an indetermined way on the momentary total dimensions of the body. 
The last equation reduces (107) to 

of which the general solution is 

/>-^/(C) ("2) 


C=logr+ / Qda (113) 

The mass within the radius r is then 

m«4;r^(C) (114) 

provided that in the function f , whose derivative is the arbitrary function 
^' in (112), the additive constant be chosen suitably, which the finiteness 
of the mass would show to be possible. 

It remains to impose the condition that the substance have a definite 
compressibility. Equations (112) and (114) yield 

and this must be a function of p only. This is equivalent to the condition 


then r""'* and r""V' niust be dependent functions of r, ^; or in terms of the 
functional determinant ^ ^^ ^, 


20 0' 

Zip' f" 


This could be obtained from the general equation (93) by substitution of 
(112) and change of variables from r, a to r, {;. It shows that, with C an 
arbitrary constant, the relation must have the form 

giving for determination of the function f 

^_3^'«C^'l (116) 

on account of which (115) gives for the equation characteristic of the sub- 

f-v *-f am 

80 that C must be negative. This is, then, the condition which it is neces- 
sary that the substance satisfy in order that the condensation under increas- 
ing mass may not be accompanied by distortion of the mass-elements. 

Conversely, if the substance satisfy this condition, the compression will 
take place without deformation. For, with the substitution of (117), equa- 
tion (6) takes the form 

The solution which is finite at the center has the form 

R^RJiiu) R^^'^p. u^R^ (119) 

where £ is a definite function, which for sufficiently small values of u can 
be expanded as a power series with numerical coefficients of alternate signs 

fl(u)=-l-BiU»+B,u*- .... 

If the radius of convergence be too limited, the function may be considered 
as determined for any value of u by analytic continuation, since equation 
(118) has no singularities to prevent. The density is then 

p -^^(u) w(u) -[fl(u)]» (120) 

so that io also is expansible for small values of u as an alternating series 
with numerical coefficients 

(jd{u) -» 1 — bju' + fcjW* — 




tn-y^(u) (121) 

where B is the definite function 




or as power-series 

Since /i is in effect a physical constant of the material, equation (121) 
shows that any particular value of u will determine a spherical suiface, 
of varying radius according to the varying value of p^, but always passing 
through the same material particles, so as to inclose the mass determined 
by that equation. If, then, u, v be two values of u corresponding to two 
definite particles, a!, ^ and o^, ^^ their distances from the center at two 
stages of the compression when the central densities are p^\ P^ respectively, 
then the definition of u gives 


a' of 

showing that the distances of two particles from the center remain in the 
same ratio as the condensation progresses. This is precisely the kind of 
contraction by which distortion of elements is avoided. 

The constant /i would be determined by the mean and surface densities 
at a given epoch from the equations 

Pi^pM^i) Pm'^^P.-r-r ^I'^P^^f^t (124) 

which would yield also the central density ^o &^ ^^^^ epoch; then for any 
epoch the first of these would determine the central density for any assigned 
total radius. 

The determination of the exact distribution of density under these con- 
ditions would rest on the computation of the coefficients so that R satisfies 
the differential equation (118); but a sufficient idea of the curve may be 
gathered simply by comparison of formula (117) with (96) and (20). All of 

these belong to the class (104), with exponents «-,-^, and 1 respectively, so 

that the first is a sort of average of the other two. With the same assigned 
mean and surface densities, it may be inferred that the hypothesis now 
considered would result in densities less in the outer strata and greater in 
the central portion than those given by the Laplacian formula (25), but 
deviating in that way to a less extent than those computed from (94). 
Such departures are opposite to the corrections which the Laplacian law 
seems to require, though the data from observation are but meager; so 
that (117) must probably be rejected, as not giving a satisfactory approx- 
imation to the actual distribution of density within the earth. The only 
satisfactory curves of class (104) would seem to be those with exponents 
somewhat larger than unity. 

The intermediate character of (117) as compared with (20) and (96) 
appears also in the phenomena which would attend unlimited accretion. 


Equations (117) and following, combined with the hydrostatic equation (3), 

du u' 
and (122) gives 


from combination of which comes 

dz z—v 

_-_J^ y.0 s^ua (126) 

This equation determines a family of curves in the y-z plane, of which the 
one required for the present purpose passes through the origin, at which 
the 2;-axis is an inflectional tangent. In the neighborhood of the origin 
this curve is given by the expansion 

with coefficients all positive. The arc required is that l3ring in the quad- 
rant where y and z are positive. » ^ 
In the half-quadrant where y>z, the inequality '■^>*^ gives y>«-, 

which shows that the curve reaches the line y^z %X some point {a,€) 
such that 0<a<^3. Crossing this line, with tangent parallel to the y- 

azis, the curve passes into the half-quadrant where y>£, where the in- 

dz €L — y 
equality -r- < ^ gives «* < a* — 2{y - a) *, which shows that the curve must 

cross the y-axis, with tangent paralld to the i-axis, at a point (z'^0,yP) 

such that 

a' 1 r- 

j9<aH — 7== or /9<a(l+2"\6)- The curvature does not change 

sign in this quadrant, since the di£ferential equation gives 

0- -l3(«-»)^-f (f-y) +«^]+«^ 

which is constantly negative where y and i are positive. This part of the 
curve is therefore a simple arch concave to the y-axis and crossing per- 
pendicularly at y-"0 and at y»)9. 

This means that there is a finite critical mass which is reached only 
when the function a)(u) becomes zero and consequently the density at the 
center infinite. The corresponding value of u is then infinite, but the 
radius is finite, since the density must be everywhere greater than p^. 

Thb result may be compared on the one hand with the case n « -^i where 

the limiting radius is reached while the mass and central density are both 
finite; on the other hand with the case n»l, where the limiting radius ii 
approached only asymptotically as the mass increases indefinitely. 

From the rejection of the hypothesis last developed it appears, there- 
fore, that under the law of compressibility which would result from any 


acceptable density-formula the compression in the interior as the mass 
increased would inevitably be accompanied by local deformations, result- 
ing from the inequality of vertical and horizontal compression, and prob- 
ably in character and magnitude similar to those described in Part I. Thus 
the theory developed from the hydrostatic equations remains subject to 
the uncertainty from this source in the computation of the compressional 
energy, to an extent not easy to estimate. 

The thermal process has thus far been outlined on two alternative 
suppositions regarding the variation of p, and two regarding a, but in all 
cases the conductivity was treated as constant. This condition might be 
replaced by the supposition often made that it increases from the surface 
toward the center on account of the increasing condensation of the mate- 
rial. Such a variation would have the effect of facilitating the transfer of 
heat from the interior to the superficial strata, probably raising the maxi- 
mum gradient of temperature attained near the surface and perhaps short- 
ening the time during which the temperature has anywhere an upward 

trend. The effect on the depth of the region where -r^ is at first positive 
is seen by inspection of equation (52), which can be written 


'ap\r^dr\ dr)^ X dr dr J 

In the second member, the first term, which alone has occurred hitherto, 
is positive in the outer region and negative in the interior; ^ is every- 

where negative, while if X increases toward the interior -r- is negative, 

making the second term positive. This shows that the zone of rising tem- 
perature would be deeper than with constant X, with the same original 
temperature curve. The more improbable supposition that X decreases 
toward the interior would have the contrary effect, but, as will be seen 
presently, could by no means eliminate the outer zone entirely. 

The fact that under a variety of suppositions regarding the thermal 
coefficients ^, a, X there occurs a thermal process marked by the same gen- 
eral features, even with no radical differences in order of magnitude in the 
numerical data, suggests that those features are not dependent on such 
special hypotheses, but due to the general properties of the original distri- 
bution of temperature and characteristic of its mode of origin. The follow- 
ing general considerations show why they may be expected to persist under 
any hypotheses on p, a, Xy not differing too radically from those developed 

Whatever may be the actual variations of pt g, o, X, any equations 
assuming to represent them as functions of r only can hardly be treated as 
more than interpolation-formulas, representing the gross features of the con- 
crete situation in the sense of averages, and disregarding the relatively trivial 
local variations on account of which it is only to a certain degree of accuracy 
that there can be said to be, for instance, a definite law of density at all. 
It seems, therefore, practically general to assume in the neighborhood of the 



surface a representation in series ascending powers of «*"ri—ri in the form 

p^Pi+a^8+€^8^+ .... g»gi+b^8+h2^+ .... 

a»ai + Ci8 + c^+ .... X'^Xi+di8+di^+ .... 

such that a few terms suffice to give all the precision which has useful mean- 
ing under the circumstances. From these come 

-^''9P'^9iPi + (9fii+pfii)9+ • • • • 
^ Pi 2ft* 





• • . • 

so that the temperature-curve in the neighborhood of the surface has the 

(?-(7s»+ C'-s^ 


and is consequently tangent to the x-axis at the surface-point :r«>l. The 
tangency is of ordinary parabolic type, since the vanishing of Oj would mean 
that the surface material was incompressible. The temperature at first 
changes at rate 

-—mm— . -- . -_Mr»--_.)=. -^ +terms with factor (r—fi) 

dt ap r* dr\ drj a^p^ ^ *' 

Since the first term here is essentially positive, there is necessarily a region 
just below the surface where the temperature rises. 

On the other hand, at the center there occur maxima of the curves for 
p, Pf and consequently for e. This would most probably happen also for 
Of so that the appropriate expansion would be of the form 

e^e.-Cr^ .... 

with which the initial rate of change is 

-rr ■■ -*+ terms with factor r* 

the first term of which is essentially negative, so that the temperature 
falls in the neighborhood of the center. The only case in which this would 
not happen would be when a would have a maximum at the center, strong 
enough to throw the maximum of temperature to a point further out, 
which is highly improbable, a necessary condition for this being that with 
expansions of the type 

/9—/9p(l— &,r* . . . .) a— ap(l — Cif* . . . .) 

the coefficients satisfy the inequality -^ ^ 



Thflse resulta do not ezdudo the oeeumnee of moio than two waomg 
alternating with fall and rise of temperature, but inqpectlon of the anal- 
ogous expansions for intermediate points seems to indicate that this would 
demand variations cS p, a, i whieh differ ^ddy in eharaeter from those 
thus far postulated, and whieh are perii^ps improbable, but whoae pcoba- 
bility it is difficult to estimate with the meager data at hand. 

Comparison of the various eases carried out in the computationB gtwm 
a f airiy definite idea of a thermal process which, from the general point 
of view hitherto adopted, can be conadered characteristic of the ptanetcs* 
imal hjrpothesb so far as. concerns independence of particular hypotheses 
respecting tibe density, conductivity, and specific heat. Under the basic 
assumptions made, the balance of evidence scans to favor its sabetantisl 
correctness in qualitative features, and even in general order of numerical 
magnitudes, since tibe chief allowances to be made in the latter can be esti- 
mated with confidence from the theory itself. For example, the irnigAa of 
time involved may be overrated, about in tibe ratio that the true oondnc* 
tivity in the interior surpasses, as it probably does, the value used, whidi 
is that obtained by observations on the rather loosely aggrq^ated Bialerial 
accessible at tibe surface. 

There seems to be sufficient reason for supposing the energy of impact 
to have little influence in determining the primitive temperature, espe* 
dally if the impinging partides be retarded serioudy by an atmoaphere. 
But adde from this tiiere have been made certain geiMral aasnmptionsb 
whieh, though fundamental in the foregoing theory, seran to be aibitraiy 
rather than essential to the generd hjrpothesis, and thus to demand critl- 
cd examination. This will be attempted in Part m, in connecticm with 
the development of an dtemative theory intended to dd in estimating the 
allowance which should be made for posdble modifications of some of these. 


As a strict theory, the foregoing deductions imply a sort of ideal earth- 
substance, with respect to which certain assumptions are made, in a form 
convenient for the purpose in hand, but not sufficient to define completely 
its thermodynamic properties. How closely they represent the actual be- 
havior of the substances composing the earth's interior is largely conjecture, 
in view of the meagemess and limited range of direct experimental informa- 
tion; but they may be examined as to their consistency with accepted 
thermodynamic laws regarding the interplay of thermal and mechanical 

It has been supposed that a general idea of the thermal process, after 
the earth was completely formed, could be obtained by treating it as a 
matter of pure conduction and accompanying radiation at the surface. 
But the significance for geological theory of the redistribution of heat lies 
largely in the resulting expansions or contractions in different portions of 
the mass, and these geometric changes would in general involve the passage 
of energy between the mechanical and the thermal form, in amount perhaps 
by no means negligible in comparison with the heat conducted. In par- 
ticular, if the temperature should on the average fall, the energy thus lost 
would be partly compensated by that developed out of gravitational work 
during the contraction. 

As to the possible relative magnitude of these, as it were, opposing move- 
ments of energy, a summary estimate may be gathered from the case of a 
homogeneous sphere at uniform temperature, contracting so as to remain 
such, and supposed to have specific heat and coefficient of expansion con- 
stant throughout. In this case the energy developed by ingathering from 
infinite dispersion is 

6 r 

while the thermal content is 
from which come 


'^^U^ S-w 


dr 6 r* d0 

where a is the volume coefficient of expansion, or three times the linear 

coefficient; so that 

_d# kma /227) 

dQ brcJ 

which is the ratio of the gain of energy from the gravitational store, to the 
loss by decline of temperature; and with an assigned density is proportional 
to the square of the linear dimensions. For small bodies it would be negli-^ 



gible; for examplOi in the ease of a planetoid a mile in diameter, composed 
of rock similar to the earth's surface strata and with a— 2X10^, it would 
be only two parts in a billion; but for a sphere having the same mass 
and dimensions as the earth, with the same value of a, it would be over 
one-fourth; in the latter case the treatment of the conduction of heat 
independently of its mechanical effects could hardly give more than a crude 
approximation. It seems conceivable that a planet considerably larger 
than the earth, even though practically solid, might exhibit the phenomenon 
described by Lane as occurring in gaseous bodies, of contraction accompanied 
by rise of temperature. 

As compared, however, with the simple case just mentioned, there is 
an essential contrast shown with a distribution of temperature like that 
described above as characteristic of the mode of origin postulated by the 
planetesimal hypothesis. Here not only are the initial temperatures near 
the surface small in comparison with those developed in the interior, but 
the changes are widely different at different depths, the interior steadily 
shrinking as the heat is conducted outwards, while the outer strata tend at 
first to expand under the rise of temperature which continues until the 
maximum surface-gradient is reached. In the early stages the surface- 
gradient is slight; the thermal energy is for the most part simply redis- 
tributed within the earth, while comparatively little is lost through the 
surface. Thus any gain of heat from potential energy on the whole could 
come only through a preponderance of the internal shrinkage. 

To estimate the nature of these movements let it be supposed that there 
is continual accommodation of the density of each portion of the mass to its 
temperature, while the accompanying variations of the pressure affecting a 
given particle are relatively negligible, so that there may be considered to 
be a definite coef&cient of expansion a, a function of r. 

If then 9 denote variations in time, the adjustment for equilibrium is 
determined by the conditions 

ddv^advdO 8v='4:7n^8r 

where 9r is the change in the central distance r of a given particle, so that 


dv^ix I adOr^dr 

The equation of expansion may also be written 

ddv ^^ a 

dv ap 

in which a is in effect the coefficient of specific volume expansion referred 
to the variation in the heat content Q instead of the temperature 0, Then 
the rate of radial motion is 

r* / dr\ dr / 

^1^1. I a^(ir'^)dr (128) 


by substitution of the value of -rr from the equation of conduction, while 

the total rate of work of gravity is 

r'dr (129) 

which by equation (3), through integration by parts, takes the alternative 
forms ^ y. 

This gives simply the total rate at which energy is transformed from the 
potential form, while the manner of its localization remains undetermined. 
The last integral, which contains only thermodynamic quantities, suggests 

that the rate of transformation per unit volume might be >l -^ ^ , but 

any such special interpretation is purely arbitrary as long as the thermo- 
dynamic substance is so incompletely defined. 

The rates of specific linear expansion horizontally and vertically are 

1 dr d /dr\ .. ,, . 

The sum of the latter and twice the former gives by (128) the rate of spe- 
cific volume expansion a -^ as it should. The moment of inertia is 

I-'^Tt I 'pT-dr (132) 

and since d[pi^dr) »0 its variation is 

which gives 

-i'L " 

dl^-^n I pi^drdr 

*-3-, v*r.i(^g)* ('») 


Any supposition as to how a varies with the depth would appear to be 
wholly gratuitous, but it may be worth while to follow out a simple one 
suggested by the form of the equations, namely, that a is constant. Under 
this condition the radial motion is 

dr ,dtf 

which is constantly negative at all depths, under all of the hypotheses en- 
tertained above in the computation of the temperature-curves, so that the 


mass everywhere shrinks toward the center, even while the temperature is 
rising in the outer strata. In particular the surface falls at a rate propor- 
tional to the temperature-gradient there, so that its shrinkage is first accel- 
erated and later retarded, the most rapid fall occurring when the tone of 
rising temperatures is disappearing. 
In this case equations (131) become 

aXdd d fsM 

The former is everywhere negative; but the latter, if ^ is constant, is nega- 
tive where the temperature-curve is convex upward and positive where it 
is concave; moreover, where €, is negative it is numerically less than €|. 
Thus the adjustment of density to temperature demands, in the interior, 
both horizontal and vertical contraction, with the former more marked; in 
the superficial strata, as long as the temperature rises, horizontal contrac- 
tion and vertical expansion. To the extent to which the mass resists defor- 
mation there are therefore developed at all depths a horizontal thrust and 
a vertical tension, which accumulate at rate proportional to €^ — e^; this 
may be written a n de\ 

Numerical or graphic differentiation from the tabulated temperatures shows 
that at any one epoch the shearing stress thus indicated is roughly propor- 
tional to the square of the distance from the center. 

The maximum possible shortening of the total radius, corresponding to 
a reduction of the temperature from its initial value down to zero through- 
out the mass is 



which reduces to 




,-4jJ .^* 


The value of ^r, between any two epochs can be found conveniently by 
comparison of this last integral with that coming from (128) integrated 
with respect to ( by substitution of the appropriate value of dO, The 

computation shows that this total shortening, with — constant, would be 

something less than 10 per cent of the whole radius; and graphic integra- 
tion from table 6 shows that about one-third of this would be accomplished 
during the epoch of rising temperatures, so that up to the time of maximum 
surface-gradient the circumference would diminish by about 800 miles. 
This number would, however, vary considerably under the different hypoth- 
eses, of which the one adopted for the moment is such as rather to exag- 
gerate the influence of the shrinkage in the central parts, and unless the 
conductivity has been seriously underrated must probably be held to refer 


to a period much longer than the whole of geological history. A measure- 
ment of the actual shortening indicated by the crumpling of the strata 
would allow an estimate of the length of time elapsed, or the " age of the 
earth. " Near the beginning the shortening is roughly proportional to the 
square of the time, and as an indication of the order of magnitude a graphic 
integration from the same table gives about 100 miles in a billion years, the 
time being, moreover, inversely proportional to the numerical value used 
for the conductivity. 

There remains to be considered what is perhaps the principal point of 
obscurity in the theory — ^the way in which the initial temperature is deter- 
mined from the work of compression. This was done above through the 
supposition that between these two there existed a definite proportionality 
indicated by the specific heat, while the latter term was not sharply defined, 
but for numerical illustration was assigned the value 0.2, an average value 
of the ordinary specific heats of certain rocks. Now if e represent the work of 

compression per unit mass, the ratio -yg for any path of compression is of 

the nature of a specific heat in physical dimensions, but its identification 
with a in any definite sense of the latter (except of course that which might 

be defined as --jg for the given path) amounts to a condition on the ther- 
modynamic properties of the "working substance" whose import there is 
need to determine. 

As concerns the relation of work and temperature, there may be con- 
sidered to be two extreme cases conceivable, illustrated by the simple 
mechanical example of a weight in frictional contact with a horizontal 
plane and drawn by a spring. If the spring is very stiff it is only slightly 
extended, and the greater part of the work of the impressed force is done 
against the friction at the area of contact; if the spring is weak, the dis- 
placement of the point of application of the force comes largely from the 
extension of the spring and the corresponding work is stored as elastic 

Corresponding to one extreme there is the fiction of a substance whose 
resistance to compression is purely frictional, its transformation of energy 
pure hysteresis — ^having at each density a certain critical pressure, the 
maximum it could sustain without further crushing, and as a function of 
the density to be used in formula (12) in computing the work of compres- 
sion. Such a substance would show no tendency to restoration of volume 
on relief of pressure; and though the manner of transformation be obscure 
it seems natural to treat the heat derived from friction during compression 
as equivalent to heat obtained by conduction or radiation, so that the ratio 
of temperature to work would depend simply on the value, at the ultimate 
density, of the specific heat in nearly the ordinary sense as related to con- 
duction at constant volume or constant pressure, at least if the coefficient 
of expansion be relatively small. 

Such an interpretation read into Parts I and II would give a more 
definite and perhaps reasonably self-consistent theory; but the conditions 
described would fail to represent the behavior of surface-rock under the 


first moderate increments of pressure, and though they might be more 
closely followed under pressures beyond the observable crushing point, yet 
the agreement, for instance, of the theoretical and empirical values of the 
modulus of compression would be little more than coincidence. 

The other extreme corresponds to what may be taken as the definition 
of a perfectly elastic substance in the thermodynamic sense, including for 
the present purpose not only fluids but perfectly elastic solids, since the 
work of shearing forces has been left out of account. Here Uie density 
depends in a definite way on temperature and pressure, independently of 
what series of changes the substance may have passed through; every path 
of change is strictly reversible, and in any closed cycle the excess of mechan- 
ical work is exactly accounted for by conduction and radiation, so that the 
work of adiabatic compression may be considered as stored elastic energy. 

The actual materials composing the earth may be judged to partake to 
some extent of the properties of both extremes. Observed cases of the 
flowage of rocks would seem to be concerned chiefly with permanent change 
of shape, with little change of volume, but it is known that the equilibrium 
of a body as large as the earth could not be purely that of an elastic solid, 
unless it should possess elastic moduli much greater than those of known 
substances, so that most probably the violent pressures occurring even at 
moderate depths would lead to some permanent diminution of volume, or 
such as partly to persist in the event of removal of the pressure. On the 
other hand, (Urect experiments on the compressibility of rocks, under what 
must here be considered small ranges of pressure, show approximately per^ 
feet elasticity, with a relatively trivial amount of hysteresis. 

It may well be that both extremes could represent acceptably the be^ 
havior of the same substance under different circumstances; for instance, 
according to the intervals of time involved. A bell made of pitch may 
sustain well-developed vibrations counted by hundreds per second and yet 
in a few hours flow into a permanently altered shape. Similarly the interior 
of the earth may be capable of sustaining seismic tremors and tidal oscilla- 
tions like an elastic solid, and yet under steady and long-continued stresses 
yield in such a way that the expenditure of energy must be counted almost 
wholly dissipative. 

Thus in view of the great length of time which must be assumed for the 
epoch of aggregation, the notion of compression with purely frictional re- 
sistance, accompanied by the production of permanent set or non-reversible 
diminution of volume, may be the appropriate one under the circumstances 
postulated by the planetesimal hypothesis, and would seem to demand no 
material modification in the essential features of the theory given in Parts 
I and II. For example, under a steadily progressive compression the pres- 
sure actually occurring with a given density at the corresponding depth 
would be at every epoch the critical pressure for that density, so that the 
density-curve for any epoch would necessarily, as was assumed without 
comment by Fisher, determine the path of compression traversed by a defi- 
nite element of the mass. 

It is likely that the phenomena of dynamical geology may themselves 
ultimately furnish the material for the most satisfactory estimation of the 


propertieB of matter here in question. Their indication seems to point dis- 
tinctly to the possibility of accumulation of truly elastic strains over periods 
of time much greater than those involved in the oscillatory movements 
commonly pointed to as witness to the existence of true elasticity. A 
reasonably complete theory would doubtless have to include the simul- 
taneous contemplation of both elasticity and viscous plasticity, of volume 
and of shape, so as to complicate the theoretical deductions enormously. 

To allow comparison it may, then, be of use for the present purpose to 
inquire what modifications are needed to give the previous theory the 
added definiteness which may come from a complete definition of the ther- 
modynamic substance, but on the supposition that this possesses the oppo- 
site extreme property, of perfect elasticity of volume under all conditions. 
There remains, of course, the same possibility as before of variety in the 
secondary features; the following developments give in some detail a 
single form as illufftration, one which has the advantage of relative sim- 
plicity in the analysis. 

Let e now represent the total intrinsic energy per unit-mass; then a per- 
fect fluid in the thermod3mamic sense, or a substance which can do work 
only through hydrostatic pressure, and has perfect volume-elasticity in the 
sense described, finds its complete description conveniently in the analytic 

form of p and e as functions of d and v<"— . The condition of conservation 
of energy as embodied in the first law ^ 

dO=|d» + (p+^)di, (134) 

gives the following determinations of auxiliary quantities 


where a, & are the specific heats, at constant volume and at constant pres- 
sure; also ap 

dv IL 

where K, H, are the isothermal and isentropic bulk-moduli and a the co- 
efficient of volume-expansion. The existence of a definite entropy-function 
c imposes the condition of integrability 

1?1„A^P^ (137) 

0'dv d0\0) 

in which case 

^.f. ^-5-Xa (138) 

dd 9 dv 60 ^ 

so that the condition of integrability is equivalent to 

I da ^ d(Ka) (I39) 

dv dO 


In addition to special hypotheses as to the form of the functioiis p, $, 
it would be necessary to specify the exact path of the compression in order 
to determine the rise of temperature produced by a given amount of me- 
chanical work. It has, however, been assumed from the beginning that in 
view of the low conductivity of rock the compression might be considered 
as relatively instantaneous and therefore adiabatic; under this condition 
the path of compression would be a curve of constant entropy, and the 
ratio of mechanical work to rise of temperature would be determined by 

de \ de p 

Id/r ^ ' . de (140) 

Comparison of this with (135) shows that -jgj^t which was treated pre- 
viously as a specific heat, can be identified with the specific heat at constant 

volume only at points where "^'O. This latter condition is satisfied 

identically by a perfect gas, which the substance might perhaps resemble 
in this respect, while di£fering widely in the relation between pressure, 
density, and temperature. 

If, however, -Tp""0 is satisfied everywhere, so that the intrinsic energy 

is a function of the temperature only, equation (137) shows that p would 
have the form p=0V and consequently by (136) that Ka*^V; where V 
is some function of v only. Since at the surface the pressure vanishes, V 
would be zero for the argument v^, hence the surface material would have 
to be either isothermally incompressible or have a zero coefficient of expan- 
sion. Now the observed compressibility and expansion of surface rock are 
enormously less than for gases, but the existence of an appreciable value 
for both of these is a necessary element in the application to d3mainical 
geology, so that a correction is called for if the above be taken as the mean- 
ing of the specific heat used in equation (58). Though there is nothing to 
impose this special interpretation, the result still suggests one way in which 
a coherent theory can be constructed, as a modification of the previous one, 
but such as to take account of the measured values of all the thermal and 
dynamical coefficients. 

Let it be supposed that the specific heat a at constant volume is a con- 
stant, understood henceforth as measured in mechanical units; that the 
intrinsic energy, instead of depending on the temperature only, has the form 

e-<7tf+^(v) (141) 

where ^(v) is a function of v to be determined; and that the isentropic lines 
have the form . 

P-/i(0-y+/a(0 (142) 


The latter condition results from the Laplacian equation (21) by treating 
h and p^ as functions of the entropy, and is suggested as a condition in view 
of the assumption hitherto made, that the path of compression is deter- 


mined by the relation between pressure and density as exhibited at various 
depths within the earth at any single epoch. For since the compression has 
been treated as adiabatic, this identification of the two thermodynamic 
paths means that as long as the effect of the conduction is insensible the 
nucleus would be in a condition of isentropic or convective equilibrium 
such as described by Bitter and Kelvin. The generalization of equation (21) 
consists in supposing that it would give the form of the pressure-density 
curve corresponding to convective equilibrium for any value of the entropy 
by proper choice of the constants h and p^. It is, then, required to deter- 
mine the functions /i,/,, ^, so as to satisfy the given conditions. 
Equation (134) shows that 

de-^jdO+^^^dv (143) 

in which the coefficient of dv must therefore be a function of v only, which 
for convenience may be written 

and then the integral of (143) is 

dip «C -•^ (• = Naperian base) ^^^) 

in which the constant of integration is considered to be absorbed in the 
undetermined function ^. 

Elimination of from (144) and (145) gives 

P-<^-^' (146) 

which must be identical with (142), and thus gives 


/iW -<^Ac+a /,(f) aBc-b (147) 


from which come 

— 1 

^-(1+Bv+c) (148) 

^^^+bv (149) 

The constant of integration in ^ is omitted, since only differences in the 
intrinsic energy are in question, so that there remain five undetermined 
constants. In tenns of these auxiliary functions, then, the properties of the 
substance are described by equations (141), (144), and (145), from which 
may be deduced the following: 


Ka~»K H -^ («Ac+a) (ISl) 

»^1+E^ (152) 


Blnee the preBsuie-deiiflity curve at the eloee of the epoeh of eompreeBon k 
•apposed to be an adiabatio line, the variation at different depths of all the 
magnitades eoneemedi for a planet of definite masBi win be determined in 
terms of the constants introduced, when a value is asrigned for the entropy, 
which may as well be taken equal to sero, since the levd of reefconing is 
arbitrary. The various surface-values are then fixed by the eondition]ig>>iOt 
with tt""0| so that the constants are subject to the c<mditionB: 

«,-4+B»i+C (154) 

5ffl2l«i._B (185) 

^-aB+fr (IM) 

If then K„Vu9u«u't ""^ eonndered to be obtained by dinet meuuze- 
ment, c and Hi may be computed from 

o-Ot'-Kivfi^* (157) 

ff^.x/l+^iE^*! (168) 

and then the wlution of (168) to (166) gives 

t>i* o 


a~?^-aA (160) 

b^^-aB (161) 

C~0t~-Bvi (162) 

leaving the single parameter A undetermined. The Laplacian constant 

*" A-H.V (163) 

in terms of which the distribution of pressure and density are still given by 
(21) and (26), while the excess of intrinsic energy and of temperature over 
their surface-values are 


*^-*^'"^» (fi-^r+^^C".-") (166) 

the former corresponding exactly to (22) above. 

In the equation (165) for the temperature the first term corresponds, 
except as to the value of the coefficient, to that which alone occurred in 
the analogous equation in Part I for the case of constant specific heat; 
while the second term is new and may be considered as the corrective term 
needed to take account of the coefficient of expansion, which occurs as one 
of its factors. The numerical value however of this corrective term is 
small; for instance, with JiC| = 4XlO^Sd|>"300 (the absolute temperature 

of the surface), e4«"2XlO~* and a^-^J, its maximum value, which occurs 

at the center where the density is greatest, is about 80^, while it contributes 
an initial surface gradient of 1^ in 14,000 meters. The value here used for 
Ki comes by comparison with the value deduced for H^ in Part I, which, as 
is seen by (158), exceeds it by only one four-hundredth part and is known 
to range well with the results of direct measurement on surface rock. 

It appears thus that to take account of all the measured mechanical- 
thermal constants relating to surface-rock it is sufficient to include only 
this trivial variation from the distribution of temperature specified by the 
first term of (165), which in algebraic form is identical with that occurring 
in the corresponding case in Part I. But the absolute value of this prin- 
cipal term depends on the coefficient A, which is here left undetermined, 
instead of having the definite value hl2a as under the former hypothesis. 
So far as concerns consistency with the conditions thus far assumed, this 
constant might be chosen so as to give any assigned value to the central 
temperature for example, or to the surface-gradient at some particular 
epoch of the conduction. But to make even the maximum surface-gradient 
match the present observed value of 1^ in 30 meters it would be necessary 
to choose A so that the initial temperatures, while similar in relative mag- 
nitude at the various depths, would be in absolute value several times as 
large as those listed in column 8 of table 2. This would perhaps be a rather 
extreme supposition, though there seem to be no definite data to the contrary. 

A suggestion from another source as to a fair mean value to be taken 
for A is the following: 

In the assumed expression (141) there are the two terms: c0, which may 
be considered for vividness as the kinetic portion, and ^(v) the potential 
portion, of the intrinsic energy e. The latter part, depending conceivably 
simply upon the mutual distances of the constituent particles, may by 
analogy with the dynamics of particles plausibly be supposed always to 
increase as v increases. Now from (149), transformed by substitution of 
the constants as determined in terms of the surface-data, comes 

,'(.)-(.A-^)(^-i,)+XM (166) 



80 that as v decreases from the value v^, the sign of ^' is ultimately that of 

H v^ 
a A ^, unless this be zero. If then ^ is to be an increasing function of 

V, the most moderate value of A which can be assumed is 

which as appears from (22) and (58) would give from the first term of (163) 
a temperature distribution precisely the same also in absolute values as 
that deduced under the first hypothesis in Part I. 

This result shows that there is not necessarily any radical difference 
between the temperature-distributions deducible from the two alternative 
extreme suppositions described, though variations in the special hypotheses 
may be expected to produce here, as under the opposite view, considerable 
differences in the quantitative conclusions. 

It has been pointed out that the surface-gradient of temperature as 
deduced from the theory would probably never reach the value at present 
found by observation, at least unless the conductivity increase consider- 
ably from the surface downward. But the present theory has taken ac- 
count only of the heat obtained from gravitational energy, to which must 
be added that obtained from several other assignable sources. These might 
be relatively unimportant, so that their effect could be included in the 
sense of corrections to the above results; but if of sufficient magnitude they 
might alter the features of the thermal process completely. 

In particular, if heat were steadily generated in each unit of volume at 
rate ^(r), the equation of conduction would be 

the solution of which may be written 

0^u+ I ^ I i^f(T)dT (169) 

where u satisfies the equation 

dii 1 d /, ,au\ 


and is to be determined so that d satisfies the given initial and boundary 
conditions. For example, if ^ is constant or the rate of generation per 
unit volume the same throughout the mass, the steady component, defined 
by the integral in (169), is, with X constant: 

whence ^ ^ 

80 that with the flux corresponding to thermal equilibrium a surface gra- 
dient of 1° in 30 meters would mean a central temperature of 100,000**, 


whatever be the values of specific heat and conductivity. An effect of such 
magnitude, produced by radio-activity or otherwise, is not to be lightly 
assumed. It is perhaps more likely that the effect of these agencies is 
largely confined to the superficial strata. 

Another possible explanation of the discrepancy respecting the tern* 
perature-gradient is that the assumed density laws may not properly repre- 
sent the variation of density near the surface. For, independently of any 
special hypothesis, it is plain that in descending from the surface the den- 
sity would tend to increase on account of the increase of pressure, and to 
decrease by virtue of the increase in temperature; which effect controls 
depends on the values of elastic modulus, coefficient of expansion, and 
thermal gradient. 

Let the density p through its dependence on p and be thereby a func- 
tion of the distance b below the surface. Then 

but since 

this may be written: 

dp ^ dp dp dp dd 
d$'' dp da 60 ds 

^P -/in ^P^ P 
de dp 

^--^(gp-Kar) (170) 

With the values /r-4X10", a«2X10-^ used above, which appear to be 
well supported by observation, this would be positive or negative accord- 
ing to whether the gradient is less than or more than 1^ in 30 meters, or 
just about the observed value. If the gradient were greater than this there 
would be at first an actual decrease of density from the surface downward, 
because of the preponderant effect of temperature over pressure; if less 
there would be an immediate increase in density such as is implied in the 
formulas (25) and (74). In the intermediate case, where Kay^gp, as ap- 
pears to be nearly realized at the present time, the density-curve would have 
a horizontal tangent at the surface-point. It would be useful if a family of 
such density-curves could be foimd, expressible in tractable analytic form, 
and satisfying the conditions named in connection with equation (92), so 
that the effect of such a modification could be determined, through a review 
of the preceding theory with the appropriate changes. 

For a discussion of the bearing of these results on geological theory; 
reference must be made to the papers of Professor Chamberlin, to whom 
the writer is deeply indebted, not only for the opportunity of co-operation 
in this work, but for suggestion and assistance freely given during repeated 





ProfenoT of ChemMtry^ Unwermiy of Chicoffo. 



When two difficultly soluble salts, such as barium sulphate and barium 
carbonate, are formed in a given medium, for instance by the addition of 
barium chloride to a mixture of potassium sulphate and carbonate, they are 
precipitated, as is well known, in the order of their insolubility. In the 
given case barium sulphate, the less soluble salt, is precipitated first, the 
more soluble carbonate last. We apply this principle in the familiar case 
of the volumetric determination of chlorides by titration with silver nitrate, 
in which potassium chromate is used as indicator; silver chromate is very 
difficultly soluble, but silver chloride is less so, and is precipitated almost 
completely and within the limits of exact quantitative analysis, before any 
solid red silver chromate can permanently be formed, the first persisting 
appearance of the latter being taken, indeed, as the evidence or indication 
that the precipitation of the chloride has just been completed. 

The application of the laws of physical chemistry, especially those of 
chemical and physical equilibrium, to such cases of precipitation shows, how- 
ever, that there must be a limit to this principle of the order of precipitation; 
the principle itself may be derived by the application of these laws and is 
always, of course, subject to them. The existence of such a limit and its rela- 
tions to the solutions in contact with given salts have been investigated 
experimentally in a number of cases, the most notable investigation being 
the classical one of Guldberg and Waage ' on the conditions of equilibrium 
between barium sulphate and carbonate, and potassium sulphate and car- 
bonate. The limiting values in all such cases are of extreme interest, as 
from them certain definite conclusions may be drawn as to the nature of 
precipitates formed in given cases; or, vice versa, from the nature of the pre- 
cipitate, conclusions as to the composition of the medium may be drawn, 
and the existence of a limiting value leads to the possibility of a complete 
reversal of the usual order of precipitation under given conditions. 

As has just been mentioned, and as will be shown in detail presently, 
the limiting value is for two given, little-soluble salts most intimately asso- 
ciated with the composition of the liquid medium from which precipitation 
occurs; the composition of the medium may in turn be dependent for one 
or more of its essential components, e.g., dissolved carbonic acid, on the 
nature of the atmosphere above the solution, with the result that the com- 
position of the atmosphere may become a fimction in the mathematical 
expressions deduced by the application of the laws of physical chemistry 
to the facts of precipitation. 

> Joumsl for Praktisehe Chemie (2), 10, 60 (1870). 



Considerations of this nature led the author in 1903/ in the course of a 
conversation with Dr. T. C. Chamberlin on the remarkable freedom from 
calcium carbonate of a deposit of gypsum/ to suggest that possibly there 
might be some connection between the purity of the gypsum and the carbon 
dioxide content of the air at the period of the gypsum formation, since the 
carbon dioxide in the atmosphere is one of the most important factors 
influencing the solubility of calcium carbonate. 

Inasmuch as climatic conditions, according to recent theories enter- 
tained qualifiedly by Dr. Chamberlin, are also dependent to a certain degree 
on the carbon dioxide in the atmosphere, it seemed possible that a study 
of the precipitation of calcium sulphate and calcium carbonate from the 
point of view of the laws of equilibrium might lead to conclusions which 
would be of some use as a further source of information relative to condi- 
tions existing at a remote period. 

Partly on account of the possibility of obtaining results of some such 
specific geological value for the work on the climates of the earth; chiefly, 
however, in order to test the possibility of exploiting this method of investi- 
gation for geological purposes, Dr. Chamberlin asked me to undertake, in 
collaboration with himself, the work of making the necessary calculations 
on the conditions of equilibrium determining the precipitation of calcium 
sulphate and calcium carbonate.* 

As is usual in these cases, it was decided to consider first the ideal case 
of solutions containing only the two salts in question. But the work has 
been extended to estimate proximately the influence of the presence of 
other sulphates in the solutions in the proportions found in ocean waters of 
the present day; and finally an attempt has been made to consider the 
effect of concentrated salt (sodium chloride) solutions on the conditions 
studied. The results obtained for the simpler case of equilibrium for dilute 
solutions in the presence of sulphates and moderate amounts of sodium 
chloride may, it is hoped, also prove useful in connection with present-day 
problems on the calcium carbonate content of the oceans and of fresh waters. 

The study of the equilibrium conditions controlling the precipitation of 
calcium carbonate and sulphate falls naturally into two parts — a study, on 
the one hand, of their relative solubilities, and the consideration, on the 
other hand, of the conditions of equilibrium for saturated solutions of cal- 
cium carbonate, calcium bicarbonate, and carbonic acid in equilibrium with 
an atmosphere of some given content of carbon dioxide, the latter being a 
determining factor in the total solubility of carbonate. This second and 
more complex part of the study will be taken up first. 

* Year Book No. 2, Carnegie Institution of Washinjgton, p. 269. 

' The observation was made by Mr. F. A. Wilder in the course of an investigation for 
a doctor's thesis submitted to Professor Chamberlin. 

* Pressure of other work made it necessary to postpone the calculations to the year 1907 



The most reliable and complete experimental determinations of the 
solubility of calcium carbonate in water containing free carbonic acid were 
made by Schloesing.^ The theoretical treatment of his results from the 
point of view of equilibrium conditions we owe to Bodlaender.' The latter's 
work, correct in its theoretical treatment/ shows an error of moment in 
the calculation of the solubility constant of calcium carbonate, the constant 
most important to us, an error due largely to an error in one of the inves- 
tigations from which Bodlaender drew his data.^ McCoy's more recent work 
on the equilibrium in aqueous solution for sodium carbonate, sodium bicar- 
bonate, and carbonic acid' gives the necessary material for the correction of 
the above error in the following pages. 

The complex conditions of equilibrium involved when water is saturated 
with calcium carbonate and with carbon dioxide under any given partial 
pressure may be developed as follows: For a saturated solution of calcium 
carbonate, say in contact with the solid carbonate, we have * 

CaCO« ^ CaCO, (1)' 

Calcium carbonate in aqueous solution is very largely ionized according 

^ CaCO, ^ Ca- +C0,^ (2) 

At a given temperature in a saturated solution of a difficultly soluble 
salt of this nature the product of the concentrations * of the ions is a con- 
stant,* which is called the solubility product or the ion product of the salt: 

CoaXCcc-Koioo, (8) 

That the product of the ion concentrations is equal to a constant for 
saturated solutions of difficultly soluble salts must at present be considered 
an empirically established fact. As is well known, the law of mass-action 
does not give constants when applied to the ionization of strong electro- 
lytes, such as salts are. For instance, for the reversible reaction 

Naa±?Na'+a^ (4) 

> Comptes RenduB. 74, 1552 (1872); 75, 70 (1872). 
'Zeitschrift fQr Physikalisehe Chemie, 35, 23 (1900). 
' See below, this page, in regard to a contested question of theory. 
« McCoy, American Chemical Journal, 29, 437 (1903). 

* Ibidem. 

* Underscoring of a symbol is used to indicate that the substance is in solid form. 

* See below in regard to the conclusion usually based on this equation that at a given 
temperature the concentration of the dissolved non-ionised calcium carbonate molecules, 
CaCO|, or tiie molecular solubility, has a constant value. 

* The term concentration, for which C is used in all the equations, is taken, in accord 
with chemical practice, to designate the number of gram molecules or moles of substance 
in 1 liter of a solution or of a gas. 

* Ostwald, Scientific Foundations of Analytical Chemistryj Nemst, Theoretical Chem- 
istry, p. 531 (1904); Bodlaender, loe. eU,; A. A. Noyes, Zeitschnft fQr Physikalisehe Chemie, 
16. 126 (1895), 26, 152 (1898), 42, 336 (1902); Report of the Congress of Arts and Science, 
vol. IV, 322 (1904); Le Blanc, Zeitschrift fOr Anorganische Chemie, 51, 181 (1906). 


we find as a matter of experience that the proportion 

is not a constant for different concentrations, as the corresponding expres- 
sion is for all other kinds of reversible reactions, including the ionixation of 
weak bases and weak acids. The above proportion for a strong electrolyte 
grows larger with increasing concentration — the most promising explana- 
tion of this apparently abnormal behavior being perhaps that the ionizing 
power of the solution is changed by the presence of considerable numbers 
of ions (the "salt-e£fect" of Arrhenius). 

This result, applied to the ionization of calcium carbonate (equation 2) , 
would mean that the proportion 

is not a constant. M)iOO» 

It is ordinarily assumed that for the saturated solutions of diflleultly 
soluble salts the molecular concentration of the salt (here Ccmx>^ u a 
constant at a given temperature in all aqueous solutions: we would expect 
then, from what has just been said, that the product Cq^XCq^qc}, could 
not have a constant value, since it is the numerator in the variable propor- 
tion (6). As a matter of experiment, however, A. A. Noyes, Findlay, and 
others found in a number of carefully studied cases of similar salts that 
the solubility products are constants, whether the given salt is present alone 
or in the presence of another salt modifying its ionization, such as a salt 
with an ion in common with it. 

The peculiar discrepancy between the empirical results expressed in (5) 
and (6) on the one hand, and the empirical result expressed in the constancy 
of the ion product on the other hand, as just explained, seems to find a 
satisfactory explanation in the work of Arrhenius on the solubility of 
salts in salt solutions.^ 

Arrhenius shows that the old view must be abandoned that in the pres- 
ence of excess of the solid salt, the solubility of the molecular or non-ionized 
salt is a constant in different saturated solutions (i.e., in solutions saturated 
in the presence of other salts). His experiments on the solubility of the 
silver salts of various organic acids (silver acetate, valerate, etc.) in the 
presence of varying amounts of the sodium salts of the same acids show 
that the solubility of the non-ionized silver salt grows smaller with increased 
concentration of the sodium salt present. In other words, there is no con- 
stant molecular solubility of a precipitate, as has so long been assumed. 
The case is analogous to the decreased solubility of gases, such as oxygen or 
carbon dioxide, in salt solutions as compared with pure water — a fact which 
is used below in the discussion of the equilibrium conditions. |^ 

Arrhenius uses his data as an argument against the correctness of some 
of Noyes's conclusions. No attempt, however, seems to have been made^by 
Arrhenius in this paper to study the ion-products for the silver salts in his 
own experiments, and calculations were therefore made with his material to 

'Zeitschrift fUr Physikalische Chemie, 31, 221 (1899). 



ascertain whether, in spite of (or rather because of) the variability of the 
molecular solubilities, the silver salts give constant solubility products or 
not. As is seen from the tables given below, the extremely interesting 
result was obtained that, according to Arrhenius's own experiments, the 
silver salts, whether present alone or with varying amounts of sodium salts 
with the acid ion in common with them, give rather good constants for 
the solubility products. 

As the question of the constancy of the solubility or ion product for a 
difficultly soluble salt is of particular importance in the investigation this 
paper treats of, space will be taken here to report the calculations made 
with Arrhenius's data for three of the silver salts, the acetate, which is the 
most soluble salt studied, the valerate, and the butyrate, which are very 
much less soluble than the acetate. 

In table 1, giving the solubility of silver acetate at 18.6° in the presence 
of varying amounts of sodium acetate, column 1 gives the molar concentra- 
tion of sodium acetate used; column 2 gives the degree of ionization of the 
sodium acetate in the mixture; column 3 gives the concentration of the 
ionized sodium acetate; column 4 the total solubility of silver acetate; 
column 5 the degree of ionization of the silver acetate; column 6 the con- 
centration of the ionized part of the silver acetate; column 7 the concentra- 
tion of the non-ionized part, which represents therefore the molecular 
solubility of the silver salt. Columns 1, 4, 5, 6, 7 are taken from Arrhenius's 
tables; the degrees of ionization of the sodium acetate as given in column 2 
were calculated with the aid of the isohydric principle, whose reliability has 
been amply demonstrated.^ Column 3 is derived from columns 1 and 2. 

For a saturated solution of silver acetate 


the solubility product would be 




in which Cgh^oo, represents the total concentration of acetate ions irrespec- 
tive of their origin from silver or sodium acetate; in each of the experi- 
ments it is the sum of the values given in a line under columns 3 and 6. The 
values of the solubility product constant K, as calculated, are given in the 
last column. 

Table 1.— ^SOvar AalaU. 















101 Mol. 


• • • • 














































* Noyes, CongreflB of Arts and Science, St. Louifl, vol. iv, p. 318, gives an excellent sum- 
mary on this question. 



Tables 2 and 3 correspond to table 1. 

Tablb 2.— Silver VaUrate. 


















• • • • 






































Tablb Z.— Silver BvtyraU. 















101 MoL 



• • • • 






























































It is quite obvious from a consideration of the values in the last column 
of each table that rather satisfactory constants are obtained in all cases for 
the ion products of the silver salts in saturated solutions. It may be added 
that of the two bracketed irregular values, the one in table 2 (last line) 
corresponds to an experiment which shows the probability of some experi- 
mental or other error in considering the seventh (next to the last) column, 
where 0.6 remains unchanged, while it should grow smaller. And the 
irregular value in table 3 (line 2) corresponds to an experiment the figures 
for which, as given by Arrhenius, show a decided divergence from the prin- 
ciple of isohydric solutions — so that it also is unreliable. 

Since, then, the results of Arrehnius's determinations, as thus calculated, 
agree with the conception that for a difficultly soluble salt the solubility 
product is a constant at a given temperature for saturated solutions, and 
since this conclusion was also reached experimentally by A. A. Noyes, 
Findlay, and others, as explained above, we may accept this now as an 
empirically-established fact. The values calculated below for the solu- 
bility product of calcium carbonate on the basis of Schloesing's data also 
show excellent agreement.* 

The value of the " solubility product " for calcium carbonate is particu- 
larly important for the study of the precipitation of calcium carbonate 
under varying conditions, and its calculation is one of the first objects of 
this investigation. Its significance lies in the fact that since in saturated 
solutions of calcium carbonate in different mixtures the concentrations of 
the calcium and the carbonate ions are variable, they need not and usually 

* Vide Stieglitz, Journal of the American Chemical Society, 30, 946 (1908), for a more 
complete discussion of this question. 


would not have equal valueSi but they are dependent on each other to the 
extent that the product of the two concentrations has a constant value for 
the saturated solutions. For instance, if to a saturated solution of the 
carbonate any acid is added, either a strong one like hydrochloric acid or 
a weak one like carbonic acid, the hydrogen ions of the acid must imme- 
diately combine to some extent with part of the carbonate ions, CO,^, to 
form more or less bicarbonate ions, HCO,': as the CO/ ions disappear, more 
molecular calcium carbonate must ionize (equation 2) to re-establish the 
equilibrium, leaving the solution undersaturated with molecular carbonate, 
and the solid carbonate, if present, must dissolve. AU this can be briefly 
expressed in the equation 

Coa X (Coo.^-«) < Koaoo. (9) 

The solution is undersaturated when the product of the concentrations 
of the calcium and carbonate ions is smaller than the solubility constant; 
similarly oversaturated (precipitation resulting) when the product is greater 
than the constant; and just saturated when the product equals the constant. 

When the solubility of calcium carbonate is thus increased by the addi- 
tion of an acid (say carbonic acid) owing to the formation of bicarbonate 
ions, HGO/, calcium bicarbonate Ga(HC03)] is formed; it remains, as most 
salts do in dilute solutions, very largely ionized. In equations (3), etc., 
Cqi^ means, of course, the total concentration of calcium ions, irrespective 
of their origin from calcium carbonate or bicarbonate. 

These relations which we have been discussing in a qualitative sense 
may be developed quantitatively as follows: 

The formation of calcium bicarbonate by the action of carbonic acid on 
calcium carbonate is a reversible reaction: 

Ca-+CO/+H+HCO/ ±? Ca- +2HC0/ (10) 


Ca- +2HC0/ ±* Ca(HCO,), (11> 

The calcium ions appearing on both sides of equation (10) in equal 
quantities evidently do not affect the equilibrium, and we have more 

^^^ ^ CO,^ + H- +HCO/ ±5 2HC0/ (12) 

and consequently: 

Coo.XCHXCHoo.-iXC?Hoo. (13) 

Canceling ChoOi o^ both sides, we have 

Cco»XCh ji-^ f^As 

— p -K'^Iooi^tioii (14) 

The study of these equations shows that, except for the formation of some 
non-ionized calcium bicarbonate, the formation of bicarbonate and the 
equilibrium between bicarbonate and carbonate are largely independent of 
the nature of the metal ion present, particularly since all salts (like calcium 
bicarbonate) are largely ionized in dilute solutions and all similar salts are 
about equally ionized in equivalent solutions. The above conclusion has 


been confirmed by the work of Bodlaender on calcium and barium bicar- 
bonate and of McCoy on sodium bicarbonate. 

Equation 14 represents the equilibrium condition between carbonate ions, 
hydrogen ions, and bicarbonate ions, or the secondary ionisation of carbonie 
acid expressed in ^^.q^, ^ ^. ^^q,, ^^gj 

and the constant of the equation may be called the second ionisation oon- 
stant of this acid. 

The primary ionization of carbonie acid, which must supply the major 
portion of hydrogen ions, is expressed in 

H,C0,±5H+HC0/ (16) 

and we have then: 

Ch X ChCO> __ XT/ /17\ 

K'loniiaUon ^^Yf then, be called the first ionization constant of carbonic 

The concentration of carbonic acid in solution is, according to Henry's 
law, at a given temperature, proportionate to the concentration or partial 
pressure of the gaseous carbon dioxide in the atmosphere with which the 
solution is in equilibrium, viz: 

CH«co.=igMXC3oo, (18) 

This completes the equations involved in the complex condition of 
equilibrium we are considering. Summarizing our conclusions, we have 
the following four mathematical equations expressing the conditions of 
equilibrium in saturated solutions of calcium carbonate and bicarbonate in 
contact with the atmosphere, all the constants of which must be simul- 
taneously fulfilled for the condition of equilibrium: 

I* Coft X Ceo, ■" Ksolub. Prod. 
II' ChsOOs " *ga8 X Ceo, 

IIP Ch X Cneo, = K'lonUatioB X Chkx), 
IV* ChX Ceo, »K^iom«ti«iX Cneo. 

The solubility of calcium carbonate, either as carbonate or as bicarbon- 
ate, reaches a limit when the solubility product of equation I is reached, 
and the value Ceo,; ^^^ concentration of the carbonate ions, in this equa- 
tion is in turn a function of equation IV; two values of this equation are 
functions of equation III, in which in turn the concentration of the dis- 
solved carbonic acid CHjeo, is dependent on the atmospheric carbon dioxide 
as expressed in equation II. 

The constants of equations II, III, and IV are known, and the constant 
of equation I, which we wish to determine, can be obtained from the other 
constants with the aid of Schloesing's experimental work on the solubility 
of calcium carbonate. 

* The solubility product of a saturated solution. 

' The solubility of carbon dioxide under varying partial pressures. 
' The primary ionization of carbonic acid. 

* The secondary ionization of carbonic acid. 


The solubility factor of carbon dioxide in water, kg„ of equation II, has 
been determined by Bunsen^ for different temperatures. The solubility is 
usually given in liters (a) of gas, reduced to 0° and 760 mm., absorbed by 
1 liter of water; then to reduce this to terms of gram molecular concentra- 
tions as required for the application of the mass-action law, we make 



since a gram molecule of a gas occupies 22.4 liters under normal conditions. 
For instance, at IQ"", 0.9753 liter is absorbed or 0.9753/22.4 « 0.04364 mole, 
if the carbon-dioxide gas has the pressure of 1 atmosphere. For any other 
pressure P of the gas, expressed in atmospheres, the molar concentration 
of dissolved carbonic acid is, then, according to II 

Ch.co,-*«mXP (20) 

In salt solutions the solubility is considerably less than in pure water, 
as determined by Setchenow* and more recently by Ge£fcken/ and, when 
salts are present, corrections made on the basis of these determinations 
wiU be used. 

The first ionisation constant of carbonic acid (equation III) has been 
determined by Walker and Comack^ from the conductivity of aqueous 
solutions of the acid. The secondary ionization is so small (see below) 
that it scarcely contributes to the conductivity of carbonic acid and con- 
sequently it can be neglected in the determination. The value found for 
the first ionisation constant is: 

K'loni^tion -3.04 X 10-' (21) 

The second ionization constant of carbonic acid (equation IV) was cal- 
culated by Bodlaender ' from Shields's * experiments on the hydrolysis of 
sodium carbonate, but, owing to an error in Shields's calculations, found by 
McCoy,' the value given by Bodlaender (1.295 X lO^'^O can not be accepted. 
It was recalculated by McCoy from Shields's results and foimd to be 12.0 X 
10~" or ten times as large as Bodlaender's value. McCoy then determined 
the constant by a study of the condition of equilibrium between sodium 
bicarbonate, carbonate, and carbonic acid, and from the values for deci- 
normal solutions of bicarbonate he obtained 6.0X10~~^", which was only 
half as large as the constant calculated from Shields's data; 0.3 normal and 
normal bicarbonate solutions gave still other values. 

A recalculation of this constant, which is important for our work, was 
made from McCoy's data, and corrections which were indicateci^^by the 
latter, but not carried out, were made for changes in the solubility of carbon 

^ liebig's Annalen, 93, 20 (1855); Dammer, Handbuoh der Anoiganisohen Chemie 
II, 372. See also Geffcken, Zeitschrift fQr PhysikaliBche Chemie, 49, 257 (1904). 
> Dammer, Handbuoh der Anorganischen Chemie, n, 1, 367. 

« Journal of the Chemical Society, London, 77, 8 (1900). 
§ Yiiui cit* 

* Zeitschrift fOr Physikalische Chemie, 12, 174 (1893). 
"^ Loe, cU, 



dioxide in salt solutions and for the changes in the ionixation of salts in 
mixtures. While the recalculation did not change the order of the results, 
it removed the uncertainty as to the effect of these corrections. 

From the two equations for the ionisation of carbonic acid. III and IV, 
we get by dividing III by IV: 

ChcO« _ K'lcmimtioB _j^ 722) 

The constant K can be readily calculated from any of McCoy's results, 
and then K^iooi«aioo 3.04 X lO"^ ,«., 

^ lonintioo^ j? " g K^) 

In 0.0999 normal solution of sodium bicarbonate according to table 2,* 
under a partial pressure of 0.00161 atmosphere (P), McCoy found 68.2 per 
cent bicarbonate and 31.8 per cent carbonate. The degrees of ionisation 
of these salts can be put equal to those of sodium acetate and sodium 
sulphate, respectively, similar salts ionising very much alike, and the deter- 
minations of the ionisation of the acetate and sulphate from their conduc- 
tivities being far more reliable than the estimations from the conductivitifiB 
of bicarbonate and carbonate solutions which really represent complex 
mixtures. A consideration of the curves for the conductivities for sodium 
carbonate and sulphate shows them in fact to be practically parallel for 
more concentrated solutions in which the hydrolysis of the carbonate is 
small, but for more dilute hydrolysed solutions the curve for the carbonate 
bends and cuts the sulphate curve, which is an indication of hydrolysis. 
Hence it was deemed safer to determine the degree of ionisation from the 
sulphate and acetate curves. Then in the experiment ' mentioned we have 

Choo, -0.0999 X 0.682 X 0.783 -0.05335 
Ceo, -0.0999X0.5X0.318X0.687 -0.01091 

The solubility of carbon dioxide in salt solutions is smaller than in 
water. A correction was made for this by putting the solubility equal to 
that of 0.1 normal potassium chloride solution. From the results of Geff- 
cken,* whose work is the best on the subject, the decrease in the coefficient 
of absorption is practically proportionate to the concentration of the salt, 
the coefficients for pure water, for 0.5 molar and 1.0 molar solutions of 
potassium chloride forming practically a rectilinear curve, the concentra- 
tions and the coefficients being used as coordinates. By interpolation on 
the curve, the coefficient of absorption for 0.1 molar potassium chloride 
at 25®, the temperature at which McCoy worked, is 0.742, reduced to 

' Loc. cU, 

'The degree of ionixation of 0.1 molar sodium acetate, Kohlrausch und Holbom, 
Leitffthigkeiten, pp. 159 and 200, is calculated as 0.783. 

The degree of ionization of 0.1 equivalent sodium sulphate, ibid,, is 0.687. In a mix- 
ture of the two, the degrees of ionization would be very slightly modified, but a calculation 
of the change made on the basis of the principle of isohyorie solutions showed the correc- 
tion to be negligible. 

' Loc. cit. 


00.760 mm. and Jb^.^-0.03313* 


Cbuco, -0.03313X0.00161 -5.33 X10-* 

Inserting the values determined for the functions in equation (22) we have 

C^HCO, „ 0.05335' ^igQ5-K 

Ch.co.XCco. 0.01991 X5.33X10-* ' *" 

The second ionization constant of carbonic acid is, then, according to 
equation (23) ^^^ 3.04X10-^ 


K^H^- :.\^. 6.21 X 10-" 

A similar calculation for experiment 6, which McCoy considered as 
presenting the most favorable conditions for accuracy and in which the 
partial pressure of carbon dioxide was 0.00404 atmosphere, gave 6.205 X 10^". 

For experiment 1, table 3,t 0.3 normal bicarbonate was used, 57.9 per 
cent remaining as bicarbonate, 42.1 per cent being converted into carbon- 
ate when equilibrium was established under a partial pressure of carbon 
dioxide of 0.00319 atmosphere. We have t && before 

Chco, -0.3X0.579X0.70 -0.1216 
Coo, -0.3X0.5X0.421 X 0.584 -0.0369 

By interpolation of Ge£fcken's results, the coefficient of absorption of car- 
bon dioxide by 0.3 normal potassium chloride at 25° is 0.720 and Ag^ ^ 
0.03214. § Hence 


C^HOO. 0.1216» ^qoiQ^K 

ChkxhXCoo, 1.02X10-^X0.0369 '''''^" 

According to equation (23) we have, then, 


We will use the mean of the two constants or 7 X lO""^^ for the calcula- 
tions made in this investigation. 



With the aid of the constant for the secondary ionization of carbonic 
acid, of the constant for its primary ionization as determined by Walker 
and Comack, and the experiments of Schloesing on the solubility of calcium 
carbonate under the influence of varying partial pressures of carbon dioxide, 

* McCoy used 0.0338 without correction for the changed solubility. 
t Page 456, he, eU. 

t The degree of ionisation of 0.3 normal sodium acetate, loe. eU,, is 0.70; of 0.3 equiva- 
lent sodium sulphate, he, eU,, it is 0.584. 
i 0.0338 for pure water. 


we can now determine the value for the solubility product of calcium 
carbonate^ n \yn xr 

to which the precipitation of calcium carbonate is subject. As explained 
above, the solution may be considered just saturated whenever the product 
of the ion concentrations is equal to this constant. Excess of carbonic 
acid increases the solubility through its hydrogen ions, which form bicar- 
bonate ions HCO3' ^^^b ^^6 carbonate ions CO/ of the calcium carbonate. 

Cdft X (CcJO,— «) < KoaOOa 

the solution is no longer saturated with calcium carbonate and the solid 
carbonate will go into solution until we again have: 

C'oa X C'cOi = KcaCO* 

the concentrations (Vof^ and G'cx>, being imequal now, calcium ions being 
in excess. 

As stated on page 242, the solubility will depend on all four of our 
fundamental equations I to IV (on p. 242), the functions being dependent 
on the various constants. They may all be combined as follows: Divid- 
ing equation III by IV, we have first: 


C^HCOa _ K lonitttUm ^ojx 

CcOs X ChsOOi K^Ioaintion 

We may substitute in this equation for Ceo, ^^ value as obtained from 
equation I, viz. ^y^ , and have 

CcaXC*HC0« ^ K^IonUation 
Ko»CO« X ChsOOs K^'ioniaatioo 


CcaXC^HCOa xT \^ ^'loniiation /ok\ 
p ='rLCmO0tXj^ (25) 

Now, when calcium carbonate is dissolved under the influence of excess 
of carbonic acid in the absence of any other calcium salts, as was the case 
in Schloesing's experiments, practically all of the calcium is present as 
bicarbonate, the quantity of carbonate being minimal and quantitatively 
negligible in comparison with the bicarbonate. Since 

Ca(HC03), ^ Ca- + 2HCO3' (26) 

the concentration of the calcium ions will be half that of the bicarbonate 

'''''^ Cca = iCHCO, (27) 

and by substituting this value in (25) we have 

C3 TTf 

IICO3 oxr vy lonisation /oo\ 

7S = ^^CaCOa X -jT^ (Zb) 

^HjCOs ^ loniaation 

For the concentration of the dissolved carbonic acid we have, according 
^°"- CH^o,°fc«.»XP (29) 

^Vide Bodlaender, loc, cit. 


where Jbg^ is expressed in molar terms and P is the partial pressure of the 
carbon dioxide above the solution (see p. 243). 
Equation (28) may be transformed then into 


2Xk^X ^J^"""^ XKcoo. (30) 

an equation which holds for saturated solutions of calcium carbonate and 
bicarbonate in equilibrium with solid calcium carbonate and gaseous car- 
bon dioxide of any pressure P at a given temperature. The composition 
of such saturated solutions has been determined by Schloesing, and since 
from his data the values of ChcOs ^^^ ^^ ^ ^^^ ^ ascertained, and since 
all the constants excepting the solubility product K(:MX)t ^^^ ^^^ known; 
the value of this constant can now be determined.'* 

The value of the first ionization constant of carbonic acid is 3.04 X 10~^;^ 
the value of the second ionization constant is 7 X 10~".^ Ge£fcken's^ recent 
very exact determinations of the absorption of carbon dioxide by waterj 
made with an improved apparatus, give somewhat higher values than found 
by Bunsen years ago. At 16^i the temperature at which Schloesing's deter- 
minations were made, by interpolation of the values found by GefFcken for 
15^ and 25°, the coefficient of absorption is found to be 0.9890 (reduced to 
0^.760 mm.) instead of 0.9753. The change in solubility produced by the 
presence of the small amount of salts present (in Schloesing's experiments 
this is only 0.01 equivalent, as an average) is negligible. The coefficient 
0.9890 corresponds to a constant Jbg^ equal to 0.04415 in molar terms. 

Substituting the values of our three known constants in equation (30) 
we obtain i /^ o /> ^ v •■ n i 

^.2X0.04415x5^^ XKo^ 

- 383.4 xKcitCO. 

Chco. (31) 


7.264 X V i" 

Schloesing's results give the total amount of calcium carbonate dissolved: 
it is present almost exclusively as calcium bicarbonate. The degrees of 
ionization, a, of the calcium bicarbonate in the various solutions may be 
put equal to the degrees of ionization of calcium acetate in equivalent con- 
centrations;^ then, in any given solut'on 

CHCOi"«XiCoft(Hoo8)j (32) 

o Bodlaender, loe. cii, 

»Page 243. 

«Page 245. 

*Loe, eU. 

'Bodlaender used calcium chloride and nitrate for this purpoee. As salta of organic 
acids usually are somewhat less ionixed than salts of stron^r morganic acids, it seemed 
better to ascertain the degree of ionixation bv comparison with calcium acetate. The con- 
ductivities for calcium acetate are ^ven in Kohlrausch and Holbom's Leitf&higkeiten on 
pase 161 and the degrees of ionization are calculated in the usual wa^. There is an average 
difference of about 4 per cent between these coefficients of ionization and those used by 



In table 4 the data are tabulated and the calculated values for ^ EQ^oOk 
given. Column 1 gives P the pressure of carbon dioxide in atmospheres, 
column 2 the solubility of calcium carbonate as determined by Schloesing 

and expressed in gram equivalents of calcium bicarbonate I — ^-^ — ^jper 

liter. Column 3 gives the degree of ionization a of the corresponding 
calcium acetate solutions, column 4 the concentration of acid carbonate 
ions, ChgOsi ^ calculated from the numb ers give n in columns 2 and 3. 

In the last column we have the values for -v Koaoot ^ calculated according 

to equation (31). 

Table 4. 






10* X 

















The mean value for y^ KckOOt is 0.002325. 

The agreement among the values found for A/KoaoOt under a partial 
pressure of carbon dioxide ranging from ^^ atmosphere to 1 atmosphere 
is excellent, and this agreement forms a very good test of the correctness of 
the whole theoretical treatment, such a constant resulting from a consider- 
ation of the conditions of equilibrium on the basis of the theory of ionisa- 

From the value for -v/Kcacos= 0.002325, we find the solubility product 

^^^^" Cca X Ceo, = Kcaco, = 1 .26 X 10"^ (33) 

It may be pointed out that this method of determining the solubility 
product constant of calcium carbonate must be far more reliable than a deter- 

' As Bodlaender (loc, cit,) points out, it is clear from the theoretical treatment that the 
same equation applies to the effect of carbon dioxide on the solubility of other difficultly 
soluble salts, sucn as barium carbonate, the only difference being that the solubility pro- 
duct constant of bariimi carbonate (Cbr X CoOs = KBaCOs) is substituted for the constant 
for calcium carbonate. All the other expressions remain the same. Bodlaender has used 
the results of Schloesing on the solubility of bariu m ca rbonate as affected by carbon diox- 
ide to calculate in the same way the value for iX KBaCOa and obtained again an excellent 

agreement among the values for the constant. Only the numerical value of the constant, 
not the constancy of the results, is affected by the corrections made in this paper, namely, 
in the value of the second ionization constant for carbonic acid and in the method of calcu- 
lation of the degrees of ionization a. As the value for v"^ KBaCOs ^ not of interest in this 
investigation, the data were not recalculated. 


mination based on the direct solubility of calcium carbonate in water. 
The latter method gives excellent results with salts b'ke silver chloride, 
calcium or barium sulphate, but calcium carbonate is largely hydrolyzed 
by water into the hydroxide and carbonic acid and bicarbonate, and this 
hydrolysis, together with the possibility of absorption of carbon dioxide 
and consequent change in the equilibrium conditions, affects these direct 
determinations. Bodlaender has estimated that about 80 per cent of the 
calcium carbonate is decomposed by water in its saturated solution in pure 
water. In Schloesing's experiments the hydrolsrsis is practically completely 
overcome by the measurements being made in the presence of an excess of 
carbon dioxide. No calculations were made by Bodlaender as to the extent 
of any hydrolysis in these experiments of Schloesing, but we can readily 
determine it as follows: In experiment 1, in which we have the smallest 
pressure of carbon dioxide and therefore the most favorable conditions for 
hydrolysis, we have as the pressure of carbon dioxide (P) 0.000504 atmos- 
phere. Consequently, according to equation (20) 


According to the table we have 

Choo,- 0.001356 

Now, for carbonic add we have (equation III, p. 242), 


and inserting the given values for Chcos ^^^ ^HiCOti ^^ ^^ • 

3.04X10-^X2.225X10-^ ^ 
^" 0.001356 -4wxnr' 

Now, for the ionisation of water at 16°, we have: 

and consequently 

0.55XlO-'« , .^.f^ 

^«" 4.99X10-^-^^^° 

Calcium bicarbonate, the chief salt in solution, is hydrolyzed according 

*^ iCa(HCO,), + HOH±* JCa(OH), + H,CO, 

and since the equivalent concentration of the bicarbonate ions in experi- 
ment 1 is 0.001356, the part hydrolyzed is 1.1 X 10~^/0.001356, or about 0.08 
per cent. So, even in this first, least favorable experiment, the hydrolysis 
is negligible. In experiment 5, we find in a similar way 

Ch=7.43x10-» Coh = 7.40X10-» 

and the part hydrolyzed is 7.4X10~*/0.0051, or 0.0014 per cent. 

Hydrolysis is reduced therefore to almost nothing, and the value found 
for Kc^ioo,, 1.26X10^, needs no correction from this source. 




For a saturated solution of gypsunii ionised according to 

CaS04±5Ca'+SO/ (34) 

we may put, as we did for calcium carbonate, 

Cca X CSO4 « KGtes64 (35) 

The value of this solubility product constant can be determined from 
the solubility of calcium sulphate in water and its degree of ionization in 
the saturated solution, since the salt dissolves without any hydrolytic de- 
composition. Kohlrausch and Rose^ give the solubility of gypsum at 18^ 
as 2.07 g. (anhydrous calcium sulphate) per liter of water. This represents 
a concentration of 2.07/136.1 or 0.0162 gram molecule and 0.0304 gram 
equivalent of calcium sulphate. The degree of ionization is best ascertained 
from its conductivity in the saturated solution; the specific conductivity 
of the solution is given ' as 0.001891 reciprocal ohms at 18^; its equivalent 
conductivity is, then 

, 0.001891X1,000 ^^^ 

'*^^* o:o3oi *^-^ 

The conductivity of calcium sulphate at extreme dilution' is 123 and its 
degree of ionization in the saturated solution therefore 62.2/123 or 50.6 per 

cent. Then Cca =Cso4- 0.0152 X 0.506 « 0.00769 


Kcd304 «= Cca X Cs04 - 0.00769* = 5.92 X lO"* (36) 

The solubility product of calcium sulphate at 18^ is therefore 5.92 X 10~*. 


If we have a solution saturated both with calcium sulphate and with 
calcium carbonate, for instance in contact with both solid salts, we have 
in the saturated solution simultaneously 

Cca X CcOs ■= KctoCOs (37) 


Cca X C!so4 = KGteso4 (38) 

The value for C(^ is the same now in both equations, representing as it 
does the total concentration of calcium ions, irrespective of their source 
from sulphate, carbonate, or bicarbonate. Then, dividing equation (38) 
by equation (37), we have for a solution in equilibrium with both salts at 

C8O4 _ Kcaso* _ 5.92X10-* _ ^ 7^ .oo\ 4 
Cc^3 " "Kcac03 " 1 .26 X 10^ " ^-^^ ^ 

'Zeitschrift fQr Phvsikalische Chemie, 12, 241 (1893). 
' KohlrauBch and Holbom, loc. cU,, p. 77. 

* Ibid., p. 200, table 86. 

* The value for KcaCOs determined at 16° is used ; no correction is made for the differ^ 
•nee of 2^. 


This means that the concentration of sulphate ions must be about 6,000 
times as large as that of the carbonate ions in a solution saturated with 
both salts, and consequently one or the other salt will come down first pure, 
when precipitation from a mixture is effected, until this ratio is reached. 
Under ordinary conditions obtaining in nature or in laboratory experiments, 
the excess of sulphate ions is not as large as this, and so the carbonate is 
precipitated first until the concentration of carbonate ions has fallen to 
about one five-thousandth that of the sulphate ions. On the other hand, 
if in such a solution, which is in equilibrium with both solid salts, the con- 
centration of the sulphate ions is by any means increased, e.g., by addition 
of some sodium or potassium sulphate, the solution will be oversaturated 
with calcium sulphate [Gq^ X G^so4 ^ ^^Q^isoj; gypsum will be precipitated, and 
owing to the loss of calcium ions the solution will now be undersaturated in 
regard to the carbonate [(Cca— «)XCoo,<Kc^ooJ> ^^^ solid calcium car- 
bonate must dissolve until the above ratio is reached. In the same way 
the order of precipitation may be reversed if by any means the concen- 
tration of the carbonate ions in a solution is persistently kept below one 
five-thousandth that of the sulphate ions during the process of precipitation 
or crystallization, for instance, by the addition of an acid.^ 

For an aqueous solution saturated with gypsum and calcium carbonate 
at about 18®, the concentration of calcium ions may be taken as 0.00769, 
as practically all of the calcium ions are derived from the sulphate. The 
concentration of carbonate ions in such a solution b, then, 

p Kciico. ^ l-26X10-* g. ^> .^Qv 

^* Coa 0.00769 == 1-^X1"^ ^*"> 

With the aid of this value we can calculate, for varying partial pressures 
of carbon dioxide, the maximum proportions of calcium carbonate and 
bicarbonate which can be present in solutions saturated with gypsum and 
calcium carbonate at approximately 18®. 



The present average partial pressure of carbon dioxide in the atmosphere 
is 0.0003 atmosphere. We may ask what is the maximum amount of cal- 
cium present as carbonate and bicarbonate in 1 liter of a solution which 
is in equilibrium with this partial pressure of carbon dioxide and which is 
saturated both with gypsum and with calcium carbonate. The significance 
of this quantity will be discussed presently; it can be calculated as folio ws,^ 
with the aid of the equations developed above. We have: 

CHXCH0Ot"=K'ioiiU»tion XChK»i (m) 

Ch X Coo, = K^Ioni-tioa X ChOO, (I V) 

GhiCOs — *gM X Ceo, (II) 

> Analogous relations have been developed experimentally and theoretieaUy in eon- 
neetion with Guldbei^ and Waage's claflsical work on barium sulphate and carbonate. 
Vid€ Nemst, Theoretical Chemistry, p. 633 (1904), and Findlay, Zeitochrift fOr Phynkalische 
Chemie, 34, 409 (1900). 


Combining the three equations by dividing m by IV and multiplying 
by 11, we find: j^, 

C?HC!0.-4;;52=^=^X*,„XCoo.XCoo. <«) 

•^ IciniMition 


Now, for a solution saturated with calciimi carbonate and gypsum at 
about 18^1 Coo, ^^7 ^ ^^ ^ & ^^ approximation to have a maximum 
value of 1.64 X 10~* as shown above (equation 40), the calcium ions produced 
from the bicarbonate being neglected in this first approximation and only 
those from the gypsum being considered.^ At 180 the solubility constant 
for carbon dioxide,' k^,^ is 0.04183 if Coot ^ expressed in atmospheres. 
In the given case Coot ^ 0.0003 atmosphere. Inserting all these values 
and the two known ionization constants of carbonic acid into equation (42), 
we find I 


Chco,- 0.0003 

For the calcium ions belonging to the bicarbonate we have 

Coa-iCHOO,- 0.00016 

We have found then the ionized portion of the calcium bicarbonate in 
the saturated solution. To determine the total dissolved bicarbonate its 
degree of ionization in the mixture must be ascertained. Its degree of 
ionization will depend not on its own concentration alone, but, according to 
the principle of isohydric solutions, also on that of the calcium sulphate 
present. We may imagine, according to the method of Arrhenius, the water 
divided between the two salts in such a way that each in its portion yields 
the same concentration of the common ion calcium. Since there is 50 times 
as much sulphate as bicarbonate, the latter will secure only about 2 per 
cent of the water, the sulphate about 98 per cent, and the sulphate will 
ionize practically as if it were present alone. Its degree of ionization is then 
50.6 per cent (p. 250), and its concentration of calcium ions 0.00769 or 
0.01538 calcium ion equivalent. This, then, must also be the concentration 
of the calcium ion equivalents in the isohydric bicarbonate solution, and so 

^® ^^"^^ C^Ca(HC0a)2Xa= 0.01538 

' The amount of calcium bicarbonate found in solution by this first approximation 
corresponds to 0.00015 gram ion of calcium. The calciiun ions from the carbonate are 
negligible and therefore the total concentration of calcium ions from sulphate and bicar- 
bonate IB 0.00769+0.00015 or 0.00784, and the maximum value for CcOa is, corrected, 
1.26XlO-«/0.00784 (equation 40) or l.iexiO"' in place of 1.64 XlO-*. No correction was 
made for this small difference, the results of the first approximation bein^ considered suiffi- 
ciently accurate, especially in view of the facts that the solubility of calcium sulphate will 
be slightly affected by the presence of the bicarbonate in such a way as to counterbalance 
this error and that the decrees of ionization of salts are uncertain. 

' Geffcken, loc. cU. The total salt concentration (0.03 mole) is too small to require a 
correction for the changed solubility of carbonic acid. 


By consideriiig the bicarbonate to ionize like the acetatCi we easily find 
by trial that ^GaCHCO,), has a concentration of 0.0196 equivalent, whose 
degree of ionization is found from the conductivity curve for calcium ace- 
tate to be 78.5 per cent 

0.0196X0.785 -"0.01538 calcium ion equivalent, 

as required by the isohydric principle.^ 

Thus we find the degree of ionization of the calcium bicarbonate to be 
78.5 per cent. The ionized portion of the bicarbonate is 0.00015 gram 
molecule per liter, as was found above; consequently the total amount of 
calcium bicarbonate in 1 liter is 0.00015/0.785 or 0.00019 gram molecule. 

A solution saturated with gypsum and calcium bicarbonate and carbon- 
ate at about 18^ under a partial pressure of 0.0003 atmosphere would contain 
therefore 1.9 X 10^ gram molecule of calcium bicarbonate. Such a solution 
by evaporation would deposit an equivalent amount of calcium carbonate, 
or 0.019 gram of calcium carbonate per liter,' and gypsum deposited by 
evaporation from such a solution would be contaminated with 0.019/2.07 
or 0.9 per cent of calcium carbonate (referred to the anhydrous gypsum).' 

For a solution saturated at 18^ with gypsum and with calcium carbon- 
ate and bicarbonate under a partial pressure of carbon dioxide 10 times as 
great as the present average value in the atmosphere, the proportion of 
calcium bicarbonate in solution can be calculated in a similar fashion. As 
a first approximation we have 

Choc - y^^^S? X 0.04183 X 0.003 X 1.64 X lO-* 


Since Cq^ is }CH00sy ^^® concentration of calcium ions in solution would 
be close to 0.00769+0.00047 or 0.00816, and consequently in the saturated 
solution the maximum value for the carbonate ions is 

p 1.26X10-^ - KAvirv-* 
^«° 0.00816 -^'g*X^^ 

Introducing this corrected value into the above equation, we find 

Choo,- 0.00916 and Coa- 0.00046 

The degree of ionization for the calcium bicarbonate is found by the 
method used above to be 78 per cent and the total calcium bicarbonate, 
ionized and non-ionized, in 1 liter, is 0.000458/0.78 or 0.00059 gram molecule. 

A solution saturated with gypsum and calcium carbonate and bicar- 
bonate at 18® under a partial pressure of 0.003 atmosphere would contain 
therefore 5.9 X 10~~^ gram molecules of calcium bicarbonate per liter, and by 

' 20 c.c. of a 0.0196 equivalent or 0.0098 molar solution of caldum bicarbonate would 
contain 0.000196 gram molecule and 0.000196X0.785-0.00015 Cca, the total concentra- 
tion of calcium ions derived from the bicarbonate in solution. The bicarbonate ionises as if 
it were all dissolved in about 20 c.c. water. 

' The amount of calcium carbonate in solution in 1 liter is so minute as to be n^ligible. 

' Cf. Findlay, loe. cU. 



evaporation such a liter would deposit 0.069 gram of calcium carbonate as 
a contamination of the gypsum or 0.059/2.07, or 2.85 per cent (referred to 
the dehydrated gypsum). Briefly the amount of calcium carbonate pro- 
duced under a tenfold increase in the partial pressure for carbon dioxide is 
very closely proportionate to the square root of 10. 

For a solution saturated at 18® with gypsum, calcium carbonate and 
bicarbonate under a partial pressure one-tenth as great as its present 
average value, viz, 0.00003 atmosphere, we may put 

Chco, = V 0. 1 X 0.0003 = 0.000096 

and Cca 0.0000475. The degree of ionization is 78.5 per cent, and the total 
dissolved bicarbonate is 0.0000475/0.785 or 0.00006 gram molecule per liter. 

A solution saturated with gypsum, calcium carbonate and bicarbonate 
at 18® under a pressure of 0.00003 atmosphere of carbon dioxide would 
contain 6 X 10~~* gram molecules of calcium bicarbonate per liter, and would 
deposit by its evaporation 0.006 gram of calcium carbonate, producing a 
contamination of 0.006/2.07 or 0.3 per cent (referred to dehydrated gypsum). 

In table 5 the results are summarized for the calcidations made for 
aqueous solutions under the conditions named. Column 1 gives the carbon 
dioxide pressure in atmospheres; column 2 gives weight in grams of calcium 
sulphate present in one liter; column 3 shows similarly the calcium bicar- 
bonate present, but expressed in grams of carbonate per liter. Ck>lumn8 
4 and 5 give the values of columns 2 and 3 in terms of gram molecules per 
liter. The last column gives the proportion of calcium carbonate which 
would be found in gypsum separating from such an ideal solution by evapo- 
ration, the proportion referring to dehydrated gypsum.* 

Tablb 5. 


(atmosp.) . 
















The significance of these results is as follows: For an ideal condition, 
if the natural waters of the earth were supposed to contain only lime salts, 
that is the sulphate, carbonate and bicarbonate in equilibrium with the 
carbon dioxide of the atmosphere, then by evaporation they would deposit 
first, as is now the case, until the solution became saturated with gypsum, 
all the calcium carbonate in solution in excess of the amounts given in 

* Cameron and Seidell, Journal of Physical Chemistry, 6,652 (1901), got experimentally 
0.037 g. calcium carbonate or 1.9 per cent from a solution containing 1.93 grams calcium 
sulphate (0.0143 mole) in equilibrium with air. The partial pressure of the carbon dioxide 
in the air is not specified (laboratory air is notably richer in carbon dioxide) and no tempera- 
ture is mentioned, hxit it probably was between 22** and 25® (see pp. 643 and 650). Only 
one determination is reported, and under the circumstances we must be content with the 
fact that the result is of the order calculated. (Vide McCoy, American Chemical Journal, 
29, 461 (1903), who also questions the quality of the air used by Cameron.) 


table 5| in columns 3 and 5, depending on the partial pressure of the atmo- 
spheric carbon dioxide. When the solution becomes saturated with gypsum 
this will, by continued evaporation, crystallize out, but no matter whether 
it is deposited in the same locality as, or in some other locality^ than, the 
first great deposit of calcium carbonate, the g3rpsum must inevitably be 
continuously contaminated with some calcium carbonate, varying from 0.3 
to 2.85 per cent, according to the partial pressure of the carbon dioxide in 
the atmosphere within the limits given. Perfectly pure gypsum would not 
be formed imder such conditions. Vice versa, a very exact determination 
of the amount of calcium carbonate present in g3rpsum prepared under 
such ideal conditions could be used as a criterion of the carbon dioxide con- 
tent of the atmosphere under which the gypsum was formed. 

Obviously such a hypothetical ideal condition as to constancy of tem- 
perature and purity of the solutions never existed on the earth, and the 
presence of other salts, notably of sulphates even in smaller quantities and 
of chlorides in larger amounts, modifies decidedly the numerical values deter- 
mined above: the foundation for the study of the relations of gypsum and 
calcium carbonate having been laid for this ideal condition, it seemed desir- 
able to pursue the inquiry to ascertain, at least roughly, the influence the 
presence of other salts would have, especially the sulphates of magnesium, 
potassium, and sodium, and the chloride of sodium. In the following pages 
an attempt has been made to estimate only roughly the influence of the 
presence of other sulphates. 



The solubility of a difficultly soluble salt like calcium sulphate depending 
on its solubility product 

Cok X Cs04 — KcaS04 

the presence of other sulphates in solutions that are not too concentrated 
to interfere with the application of the laws of solution would, through the 
increase in the concentration of the sulphate ions, have as its chief effect a 
decrease in the concentration of the calcium ions, according to the equation 
just given. Since about half of the gypsum is ionized in its saturated solu- 
tion, a decrease in the concentration of its calcium ions would imply a 
decrease in the solubility of gypsum. The decreased solubility of the gyp- 
sum or the decrease in the concentration of calcium ions would, vice versa, 
increase the solubility of calcium carbonate and bicarbonate; so, from both 
causes, gypsum crystallizing from a not too concentrated sulphate solution 
under the conditions we are stud3ring might be expected to be more contami- 
nated with calcium carbonate than was found for aqueous solutions. 

In order to study these effects, it was decided to examine from this 
point of view the probable effect of the sulphates of magnesium, potassium, 
and sodium in the proportion in which they are found with calcium sul- 
phate in the present sea-water, determining also the effect of var3ring partial 
pressures of atmospheric carbon dioxide, as was done before. 

' Such a change of locality was stiggeeted to the author by Dr. O. Willcox. 


At present we have in sea-water in 1,000 parts 1.239 grams or 0.009 
gram molecule of calcium sulphate, 1.617 grams or 0.0135 gram molecule 
of magnesium sulphate, and 0.860 gram or 0.005 gram molecule of potassium 
sulphate.* Such a solution would contain 0.0275 mole or 0.055 gram equiv- 
alent of total sulphates. The degrees of ionization of calcium and magnesium 
sulphates are practically the same,' and while potassium sulphate ionises 
far more readily, it forms a relatively small component of the system, and 
we can, with sufficient accuracy for our purpose, consider the degree of 
ionization for the sulphates to be that of a 0.05 equivalent magnesium sul- 
phate solution or 48 per cent. Then 

qso4 - 0.0275 X0.48 - 0.0132 

Go»-0.0091 X0.48:-0.0044 

CoaXCso*- 0.000058 

Since for a saturated solution the solubility product constant is 0.000059 
(EcuaoJ/ ^6 s^ that this sea-water would be almost saturated with gyp- 
sum were it not for the modifying influence of the presence of a large pro- 
portion of sodium chloride and other salts in it.^ 

A purely aqueous solution of the above composition should be nearly 
saturated with gypsum and a large part of the latter should crjrstallize out 
during its concentration say to one-quarter its original volume. If at the 
beginning, when the solution is practically saturated with gypsum, it also 
at the same time be considered to be saturated with calcium carbonate and 
bicarbonate in equilibrium with a partial pressure of 0.0003 atmosphere 
carbon dioxide, we would have (see p. 250) a maximum concentration of 
carbonate ions q, q q^qo 

^^" i;7oo " i;7oo""^-* ^ ^°^ 

Then, according to equation 42 (p. 252) we must have sufficient calcium 
bicarbonate in solution to give a concentration of acid carbonate ions 


Ch(X). - V ^^^^3^=^ X 0.0003 X 0.0^^ 


and the concentration of the ionized calcium carbonate would be 0.000195 
gram molecule. Its degree of ionization is 74 per cent, calculated by the 
method used before, so the total calcium bicarbonate in solution is 0.000264 
gram molecule per liter. 

* Ghamberlin and SaliBbury, Geology, p. 309. 

' Kohlrausch and Holbora, loe. eit.f p. 200. For 0.01 equivalent the degrees of ionisa- 
tion are 63 and 65 per cent respectively. 

* Page 250. 

* In this calculation we are not including a study of the effect of sodium chloride, 
but are limiting ourselves to an examination of the effect of sulphates alone in an aqueous 
solution. It is imderstood that the calculations are only for a rough orientation, the sim- 
plicity of the laws for dilute solutions being lost as solutions become more concentrated and 
more complex. Vide also E. C. Sullivan, Journal of the American Chemical Society. 27, 


If such a solution is now allowed to evaporate under a partial pressure 
of 0.0003 atmosphere of carbon dioxide to one-quarter its original concen- 
tration, we should have left a solution containing approximately 0.08 gram 
molecule of sulphates per liter, whose degree of ionisation may be taken as 

39 per cent.* Then Qgo^^ 0.08X0.39 -0.0312 

and the maximum concentration of carbonate ions would be 




and the concentration of ionised calcium bicarbonate is 0.0003 mole. Its 
degree of ionization is 63.4 per cent and the total concentration of calcium 
bicarbonate 0.00047 mole. Comparing this result with the original solu- 
bility of the bicarbonate (0.000264 mole), we find that while the solubility 
of the gypsum is decreased (see below) by the accumulation of the other 
sulphates, the solubility of the bicarbonate increases, in consequence of the 
decreasing concentration of the calcium ions. 

The total original gypsum in solution was 0.0091 gram molecule per 
liter. By evaporation of the solution to one-quarter of its volume, C8O4 
has become 0.0312, and consequently 

p 0.000059 

^" 0.0312 "°°°^® 

which we may consider as derived entirely from the ionization of the gypsum. 

Its degree of ionization in a 0.08 molar solution of sulphates is about 39 
per cent, and therefore the total concentration of the gypsum left in solution 
is about 0.0019/0.39 or 0.0049 gram molecule. A quarter of a liter would 
contain 0.0012 gram molecule, and therefore of the total original 0.0091 
gram molecule 0.0079 gram molecule or about 90 per cent would have 
crystallized out. The weight of this would be 1.07 grams (calculated as 
calcium sulphate).' 

The total original concentration of calcium bicarbonate was 0.000264 
gram molecule, and we have left a quarter of a liter with 0.00047/4 gram 
molecule. Consequently, 0.000147 gram molecule or 0.0147 gram of cal- 
cium carbonate should be deposited with 1.07 grams of calcium sulphate, 
corresponding to a contamination of 1.37 per cent. 

A further evaporation to one-eighth of the original volume should de- 
posit according to similar calculations 0.101 gram calcium sulphate with 
0.0047 gram calcium carbonate, representing 4.7 per cent. 

With a partial pressure of 0.00003 atmosphere of carbon dioxide the 
original solubility of the bicarbonate would be 8.3 X 10~~* moles. By concen- 
tration of 1 liter to one-quarter of a liter, the solubility would be increased 

* The degree of ionization of magnesium sulphate, the chief sulphate left. 
> During the evaporation to one^uJf liter, we should expect a deposit of 0.72 gram sul- 
phate with 0.0104 gram carbonate or 1.44 per cent, carbonate would be present. 



to 14.8 X 10~* gram molecule per liter and only 0.0046 gram of calcium car- 
bonate would be deposited with 1.07 grams of calcium sulphate crystalliz- 
ing out as gypsum by the concentration! representing a contamination of 
0.43 per cent. 

With a partial pressure of 0.003 atmosphere of carbon dioxide the 
original solubility of calcium bicarbonate would be 10 times as great as in 
the case just discussed, amounting therefore to 8.3 X 10^ moles. Evapora- 
tion of the solution to one-quarter its volume would increase the solubility 
to 14.8 X 10~^ gram molecules per liter and 0.046 gram of calcium carbonate 
would be deposited with 1.07 grams of sulphate, representing 4.3 per cent. 

The results are summarized in table 6.^ 

Tablb 6. 















Column 1 gives the carbon-dioxide pressure in atmospheres; column 2 
gives the weight in grams of calcium sulphate in 1 liter; column 3 the solu- 
bility of calcium carbonate and bicarbonate (expressed in grains of calcium 
carbonate) ; columns 4 and 5 give the values of the two previous columns 
in moles per liter. The last column gives the contamination of gypsum with 
sodium carbonate (referred to dehydrated gypsum) produced by evapora- 
tion of three-quarters of the solution. 

This table can be regarded only as a rough approximation, as it was 
not considered desirable to make more involved calculations at the present 
time, and certain factors modifying the results were therefore not considered, 
such as the change in solubility of carbon dioxide, the influence of the 
primary ionization of potassium sulphate, etc. It suffices, however, as an 
orientation as to the effect of the presence of sulphates on the probable 
contamination of gypsum with calcium carbonate: the decreased solu- 
bility of gypsum in the presence of such sulphates, allowing the calcium 
bicarbonate to dissolve more nearly as it does in the absence of gypsum, 
would increase the tendency to contamination of the gypsum and so a 
perfectly pure or exceedingly pure gypsum crystallizing under such condi- 
tions would a fortiori indicate a very low partial pressure of carbon dioxide. 

* The calculations involve no greater total concentration than about 0.08 gram mole- 
cule total sulphates per liter (p. 257). Cameron and Seidell, Journal of Physical Chemis- 
try, 5, 650 (1901), have determmed the solubility of gypsum in the presence of sodiimi sul- 
phate and found that 0.1 gram molecule sodium sulphate at 22^ reduces its solubility from 
2.1 grams calcium sulphate to 1.4 grams, which is qualitatively in accordance with the 
above calculations. No attempt was made by them or by the author to determine to what 
extent the ion product for calcium sulphate remains a constant in these determinations. 
In more concentrated solutions of the sodium sulphate, as Cameron and Seidell point out, 
double salt formation increases the solubility of the gypsum. 




In both the previous cases discussed it was found that the absence of 
calcium carbonate in gypsum crystallizing from solutions containing cal- 
cium carbonate in equilibrium with the carbon dioxide of the air would be 
considered an indication of a very low partial pressure of the carbon dioxide. 
Even if the great mass of the excess of calcium carbonate in solution were 
deposited first, in the same locality or elsewhere, before the point of satura- 
tion for gypsum were reached, the requirements for equilibrium would be 
such as to keep so much of the carbonate in solution as to form an easily 
discernible contamination of the gypsum formed by further concentration. 

In nature the crystallization of gypsum is supposed to occur usually by 
the concentration of waters containing a large excess of other salts, notably 
of sodium chloride, and the last question we shall try to consider now is the 
effect of sodium chloride on the conditions discussed in the preceding parts. 

A salt like sodium chloride which has no ion in common with gypsum 
should, according to the law of mass-action, increase the solubility of the 
latter up to a certain point; the chloride and sulphate must react to a con- 
siderable extent to form calcium chloride and sodium sulphate: 

CaS04-l-2Naa ±5 Caaj + NajSO^ 

As the ionization of the new salts in moderately concentrated solutions is 

by no means complete, considerable amounts of calcium and sulphate ions 

must be suppressed to form these salts in non-ionized form, and this would 

lead to an increased solubility of calcium sulphate according to equation 

(35), page 250. In fact, a rough calculation of the result of Cameron's ' 

determination of the effect of sodium chloride in 0.017 molar solution on 

the solubility of gypsum, made with the aid of Arrhenius's principle of 

isohydric solutions, led to a value for the ion or solubility product for 

calcium sulphate 


in the presence of the salt, as compared with 6.5 X 10~^, the value of the solu- 
bility product at 23® in the absence of salt. 

For concentrated salt solutions the conditions, as is always the case, 
become more and more complex; we are more likely to have complex ions, 
such as NaS04, and double salts formed in large quantities and leading 
not so much to abnormal changes in solubility as to changes which we have 
no means of estimating at present.' Cameron, however, has given us a 
large amount of empirical data which will be useful for the consideration of 
our subject. In regard to gypsum and sodium chloride^ he finds the solu- 
bility of the former is rapidly increased, rising from 2.37 grams calcium 
sulphate (0.0174 mole) per liter to 7.50 (0.0555 mole) in the presence of 
130 grams of sodium chloride, after which there is a gradual decrease in 

* Loc, cU,, 5, 660. 

' As to the lowering of Bolubilitv of the non-ionised calciiun sulphate in salt solutions, 
comparable with the decreased solubility of carbon dioxide, see pp. 237, etc. 




solubility again, which Cameron ascribes to the condensation of the solvent 
in aqueous solutions of electrolytes.^ 

Entirely analogous effects must be anticipated for the action of sodium 
chloride on calcium carbonate and bicarbonate, their ions also being sup- 
pressed, for instance, according to 


calcium chloride and sodium chloride being formed in considerable quanti- 
ties according to the principle of isohydric solutions.' 

Again, the molecular solubility of calcium carbonate is liable to be 
decreased, as was discussed for the sulphate, and we have in this case also 
the fact that the solubility of carbon dioxide is considerably less in salt 
solutions than in pure water, and this decreased solubility will reduce the 
amount of bicarbonate dissolved approximately proportionately to the cube 
roots of the change in the coefficients of absorption.' So we have forces 
tending to increase the solubility as well as such as tend to decrease the 
solubility of calcium carbonate. As a matter of fact, Cameron found that 
at 25® for calcium carbonate and bicarbonate in equilibrium with air,^ the 
solubility was increased from 0.1046 gram bicarbonate per liter to 0.2252 
gram by 51 grams (0.87 mole) sodium chloride, and then it was decreased by 
additional sodium chloride. There is, therefore, at first an increase in solu- 
bility and then a decrease, exactly as for gypsum. The characteristic bend in 
the curve occurs earlier than in the case of gypsum, which was to be expected, 
as the molecular solubility of carbon dioxide is also affected in this case. 

Cameron * also determined the effect of sodium chloride on the solu- 
bilities of calcium sulphate, carbonate, and bicarbonate simultaneously, 
i.e., in mixtures in equilibrium with solid gypsum, solid calcium carbonate, 
and the air. This is the work that is of most interest and importance for us, 
and as we shall use the data the table is reproduced here (table 7). 

Table 7. 











(molea). | 













0.0625 1 







0.1979 1 







0.6824 1 






























^ hoc. cit.j by 576. 

' This has been shown to be reliable up to 0.4 normal solutions. (Summary by A. A. 
Noyes, loc. cU,) 

' When no other calciiun salt is present, the solubility changes according to the cube 
root of the concentration of the carbon dioxide (equation (30), p. 247). When another 
calcium salt, e.g., gypsum, is present in large excess, so that Cca may be considered constant, 
the solubility changes approximately as the square root of the concentration of carbon 
dioxide (equation (25), p. 246, and (42), p. 252). 

^ No mention is made as to the partial pressure of the carbon dioxide, whether *' labora- 
tory air" or pure country air was used. 

» hoc. cU,, 5, 653. 



We note, first, the interesting fact that the solubility of the calcium 
bicarbonate is very much reduced by the presence of the gypsum, as required 
by the theory developed above. Cameron did not report any experiments 
on the efifect of a change in the partial pressure of carbon dioxide, but it 
will undoubtedly be entirely analogous to that discussed in the previous 
parts of this paper; it is probable that the solubility will increase roughly 
proportionately to the square root of the pressure of carbon dioxide.^ 

We may now raise the question whether, owing to the displacement of 
the bend in the curve of the solubility of bicarbonate as compared with 
that of gypsum in the presence of sodium chloride, by a process of concen- 
tration gypsum would crystallize out of any of the above solutions free 
from carbonate, and if not what the contamination would be. 

If we go from solution 2 to solution 3, the sodium chloride concentration 
is increased from 0.0625 to 0.1979 mole, which would result if 100 liters of 
solution 2 were concentrated to 31.54 liters of solution 3. Of the 272 grams 
of calcium sulphate in 100 liters of solution 2, 108.7 grams would be retained 
in solution and 163.3 grains deposited. At the same time, of the 4.5 grams 
calcium carbonate in solution (as bicarbonate), 2.77 grams would be de- 
posited, that is 1.7 per cent referred to pure calcium sulphate. 

Table 8 gives the results of the calculations of this efifect of progressive 
concentration from solution 2, through solutions 3, 4, etc., to solution 8. 
In column 1 the number of the solution used is given, referred to the num- 
bers in table 7. In column 2 the number of liters used for concentrating is 
given, and in the next column we have the number of liters to which the 
solution has been concentrated. Column 4 gives the weight in grams of 
ealcium sulphate deposited, and the last column gives the proportion of 
calcium carbonate to anhydrous sulphate deposited. 

Tabls 8. 











No. 2 





No. 3 





No. 4 





No. 6 





No. 6 





No. 7 





We see from table 8 that the most favorable point for the deposit of 
pure gypsum is in going from solution 5 to solution 6, which is the point 
where the curve of solubility of gypsum bends, its solubility now decreas- 
ing and a relatively large amount of it being deposited; the bend in the 
curve of the solubility of the bicarbonate has already been passed and there 
is now no corresponding increase in the formation of carbonate, this increase 
having occurred before and produced the strong contamination in the 
previous series. 

^ See note 3, p. 260. 


Nothing is stated in Cameron's paper whether the air used was "labo- 
ratory air'' with an excess of carbon dioxide in it or pure country air.^ If 
laboratory air was used, the amount of calcium bicarbonate obtained in 
solution and subsequently precipitated would be excessive and the results 
in the last column of table 8 might be materially lower for air with a carbon- 
dioxide content of 0.0003 atmosphere. 

It is interesting to note that according to Usiglio's ' work on Mediter- 
ranean water, calcium sulphate began to be deposited when the water 
reached a density of 1.13. This corresponds to a chloride content of about 
17 per cent, or about 3 gram molecules of sodium chloride per liter, a con- 
centration reached in Cameron's experiments for solution 7, from which, 
going to solution 8, gypsum would be obtained with about 0.8 per cent car- 
bonate. At this concentration, the concentration of other sulphates is 
still so small that they would tend to increase the contamination with 
carbonate, as described above, rendering the calcium sulphate less soluble 
and the bicarbonate more so. 

At a lower partial pressure of carbon dioxide than 0.0003 atmosphere, 
the proportion of carbonate would be reduced approximately in the ratio of 
the square roots of the ratios of the partial pressures,* i,e,, a partial pressure 
of only 0.00003 atmosphere would reduce the carbonate to about ^ parts of 
the values given in the last column of table 8, or to still less on the prob- 
able assumption that the experiments on which the table is based were not 
carried out with pure air, and would produce a very pure deposit. 


Reliable data on the solubility of calcium carbonate at temperatures 
other than 16^, the temperature at which Schloesing's experiments wer& 
carried out, are not at hand, and so the effect of changes of temperature om 
the conditions we are studying can not be estimated. It may be pointed out, 
however, that aside from a probable increase in the solubility of calcium 
carbonate, a higher temperature would affect chiefly the solubility coefli- 
cient of carbon dioxide, and through it would reduce the formation of 
calcium bicarbonate. The solubility * of gypsum is about the same at 50** 
to 65^ as at 18^ and ionization constants are usually not changed greatly 
by changes of temperature; but the coefficient of absorption * for carbon 
dioxide at 65.5^ is just about one-ninth as large as the coefficient at 18^. 
This would result, according to equation (42), in reducing the formation of 

calcium bicarbonate to -J i or one-third of the value found for 16®. In other 

words, a rise of temperature of some 50® would probably have about the 
same effect on the solubility of calcium bicarbonate as a decrease in the 

* Results obtained by McCoy (American Chemical Journal, 29, 461 (1903), in repeating 
other work of Cameron on conditions of equilibrium involving the carbon dioxide of the 
air and showing decided discrepancies, form a very strong indication that pure air was not 
used by Cameron. 

* Encyclopsedia Britannica. 
» See note 3, p. 260. 

* Comey, Dictionary of Solubilities, p. 422. 

^ Dammer, Handbuch der Anorganischen Chemie, n, 1, p. 371. 


partial presBure of carbon dioxide to one-ninth its present value. In view, 
howeveTi of the unknown change of solubility of calcium carbonate (the 
ehange in the ion product constant), it must remain undecided whether a 
rise of temperature would be a favorable or an unfavorable factor in the 
crystallisation of gypsum free from carbonate. The formation of bicarbon- 
ate would also be directly proportionate to the square root of the solubility 
eonstant according to equations (40) and (42). 


(1) From Arrhenius's data on the solubility of silver acetate, valerate, 
and butyrate in the presence of the sodium salts of the same acids, it was 
shown that the solubility or ion products are approximate constants, as 
calculated on the basis of the well-established principle of isohydric solu- 
tions. This, with the results of others,^ removes the discrepancy existing 
in the relation between the solubility product and the ionization of strong 
electrolytes and gives us a safer empirical foundation for the consideration 
of the equilibrium conditions existing between two precipitates, one which 
is in harmony with the fundamental work of Guldberg and Waage. 

(2) The second ionization constant of carbonic acid may be taken as 
7.0 X 10^'^ as calculated from McCoy's data with the aid of corrections sug- 
gested but not carried out by the latter. 

(3) The solubility product of calcium carbpnate is found to be 1.26 X 10~* 
on the basis of Schloesing's experiments on the solubility of calcium car- 
bonate at 16^ under varying partial pressures of carbon dioxide. 

(4) The solubility of calcium carbonate and calcium bicarbonate is ap- 
proximately proportionate to the square root of the partial pressure of 
carbon dioxide in the presence of a large excess of calcium sulphate. 

(5) The theory of the equilibrium conditions between calcium sulphate 
and calcium carbonate and bicarbonate has been developed and may prove 
useful in the study of the natural waters of the present day. 

(6) Considering the results given in tables 5,' 6,' and 8,^ we find that the 
favorable factors for the crystallization of pure gypsum should be: 

(a) The absence of other sulphates which in moderate proportions 
render gypsum less soluble and consequently enable solutions to take up 
more calcium bicarbonate than pure aqueous saturated solutions of gypsum 
can dissolve. 

(b) The presence of sodium chloride in the proportion of about 8 to 25 
per cent.* 

(c) A very low partial pressure of carbon dioxide, the solubility of cal- 
cium carbonate varying approximately as the square root of the partial 
pressure of carbon dioxide, according to (4). 

(d) An increase of temperature by decreasing the coefficient of absorp- 
tion of carbon dioxide would possibly, but not certainly, be a favorable fac- 
tor, the formation of calcium bicarbonate being proportionate to the square 
root of the absorption coefficient of carbon dioxide, which falls with increase 
of temperature. 

> A. A. Xoyes, loc. cU,; Findlay, loc. eU. < Page 254. < Page 258. « Page 261. 
* Solutions 5, 6, 7, table 7, p. 260. 

■ > 
- «• 


It is conceivable that the conditions (a), (b), (c), and (d) should lead to 
a primary deposit of exceptionally pure gypsum, especially when acting 
jointly. The considerations developed make it desirable to. examine such 
and other deposits of gypsum very carefully and exactly for even very 
small quantities of carbonate. 

(7) Even if the great mass of an excess of calcium carbonate in a solu- 
tion were deposited first in some other locality before the point of satura- 
tion for gypsum were reached, the requirements for equilibrium would be 
such as to hold carbonate in solution and to make the question of the place 
of deposit of the excess of carbonate in the first instance one of no moment. 






i .' 

I f 


i ,1