18 Prof. G. II .
Darwin .
[ Nov. 22 , November 22 , 1883 .
Professor T. H. HUXLEY , President , in the Chair .
In pursuance of the Statutes , notice was given from the Chair of* the ensuing Anniversary Meeting , and the list of Officers and Council nominated for election , was read as follows:\#151 ; President.\#151 ; Professor Thomas Henry Huxley , LL. D. Treasurer.\#151 ; John Evans , D.C.L. , LL. D. Rprrpfnrip* / Professor George Gabriel Stokes , M.A. , D.C.L. , LL. D. ' l Professor Michael Foster , M.A. , M.D. Foreign Secretary.-\#151 ; Professor Alexander William Williamson , LL. D. Other Members of the Council.\#151 ; Captain W. de Wiveleslie Abney , R.E. ; Professor W. Grylls Adams , M.A. , F.C.P.S. ; the Duke of Argyll , K.T. , D.C.L. ; John Gilbert Baker , F.L.S. ; Thomas Lauder Brunton , M.D. , Sc. D. ; William Henry M. Christie , Astron. Royal ; Warren De La Rue , M.A. , D.C.L. ; Sir Frederick J. O. Evans , K.C.B. Professor George Carey Foster , B.A. ; Francis Galton , M.A. , F.G.S. James Whitbread Lee Glaisher , M.A. ; Sir William Withey Gull , Bart. , M.D. ; Hugo Muller , Ph. D. ; Professor Joseph Prestwieh , , M.A. , F.G.S. ; Professor Osborne Reynolds , M.A. ; Osbert Salvin , M.A. , F.L.S. The Presents received were laid on the table , and thanks ordered for them .
The following Papers were read:\#151 ; I. " On the Formation of Ripple-mark in Sand .
" By G. H. Darwin , F.R.S. , Plumian Professor and Fellow of Trinity College , Cambridge .
Received October 18 , 1883 .
The following paper contains an account of experiments and observations on the formation of ripple-mark in sand .
The first section ia devoted to experiments on the general conditions under which ripple-mark is formed , and especially on the mode of formation and maintenance of irregular ripples by currents .
In the second section it is-shown that regular ripple-mark in sand is due to a complex arrangement of vortices in oscillating water ; and the last section gives some account of the views of certain recent observers in this field , and a 1833 .
] On the Formation of Ripple-mark .
discussion of some phenomena in the vortex motion of air and water .
S 1 .
First Series of Experiments .
A cylindrical zinc vessel , like a flat bath , with upright sides , 2 feet 8 inches in diameter and 9 inches deep , was placed on a table , which was free to turn about a vertical axis .
Some fine sand was strewn over the bottom to a depth of about an inch , and water was poured in until it stood three inches deep over the sand .
After some trials of simply whirling the bath , in which no regular ripple-mark was formed , I found that rotational oscillation with a jerking motion of small amplitude gave rise almost immediately to beautiful radial ripples all round the bath .
If the jerks were of small amplitude the ripples were small , and if larger they were larger .
On one occasion having made large ripple-marks , I oscillated the bath much more rapidly , and a second set of ripples sprang into existence in the furrows of the first set .
Another time , when in consequence of .
irregularity in the motion , a set of radiating waves were generated in the water , a second set of transverse ripples were formed , which produced by interference a beautifully mamellated structure , arranged like a chess-board .
In all these experiments the radiating ripples began first to appear at the outer margin of the bath and grew inwards ; but the growth stopped after they had extended to a certain distance .
If the jerking motion was violent , ripples were not formed near the circumference , and they only began at some distance inwards .
After these preliminary trials , arrangements were made for regularising both the frequency and amplitude of oscillation of the bath .
An attempt was then made to formulate the laws which govern the generation of ripple-marks .
In the following notes of experiments the expression " octave " is used to denote a ripple-length which is one-half of the main or fundamental ripple , and the amplitudes are measured by the displacement of the edge of the bath .
The water stood 1 inch deep in the bath .
1 .
Amplitude 1 inch ; frequency 52 per minute ( complete oscillations ) .
No ripples formed after four minutes .
2 .
Amplitude 2^ inches ; frequency 52 .
55 ripples round the circumference , extending about 4 inches inwards ; somewhat irregular and with a tendency to break into the octave .
3 .
Amplitude 6-g- inches ; frequency 52 .
Motion very violent .
About 5 large irregular ripples in the circumference , breaking at about 9 inches from the outside into about 40 ripples .
4 .
Amplitude 1^ inches ; frequency 52 .
Prof. G. H. Darwin .
[ Nov. 22 , In one part 7 ripples to 9 inches ; at another part 12 ripples to 1 foot ; some tendency to break into the octave .
The ripples only extended an inch or two from the edge .
More water was then poured in until it stood about If inches deep .
5 .
Repetition of No. 4 .
There were 33 ripples in a half circumference ( which is 22 inches ) ; the ripples were more regular than in No. 4 , with not so much tendency to break into the octave .
6 .
Water 2\ inches deep ; frequency 59 ; amplitude 3 or 4 inches .
43 ripples to the circumference ( which is 44 inches ) , extending inwards 8 inches .
Hereafter the amplitudes were marked by a pointer projecting 3 inches from the edge of the bath .
7 .
Water 2 \inches ; frequency 75 ; amplitude inches .
66 ripples to circumference , very regular , and extending inwards 4 or \amp ; inches .
8 .
Water 2^ inches ; frequency 74 ; amplitude 3 inches .
63 or 64 ripples , extending 6 or 7 inches ; broken in two or three places .
9 .
Water 2j % ; frequency 75 ; amplitude 4 inches .
53 ripples , extending 8 or 9 inches ; not so regular .
10 .
Water 2\ ; frequency 78 ; amplitude 5 inches .
47 ripples , extending 10 or 11 inches .
11 .
Water 2\ ; frequency 75 ; amplitude 6 inches .
Agitation violent ; all the coarser sand collected round the margin , without ripple-mark for 4 inches inwards ; from 4 to 11 inches inwards , rather irregular ripples about 37 to circumference ; the usual flat centre .
12 .
Water 2\ ; frequency about 80 ; amplitude 7 inches .
The water churned up the sand with violence ; margin the same as No. 11 ; from 8 to 12 inches from margin rather irregular ripples , about 34 to circumference .
13 .
( Bad observation .
) Water 2\ ; frequency about 85 ; amplitude about 1-^ to 2 inches .
80 ripples to circumference .
14 .
( Bad observation .
) Water 2\ ; frequency 57 ; amplitude 2 inches .
No ripples raised .
An analysis of the observations marked 7 to 14 was made on the hypothesis that the water remained still , when the bath oscillated with a simple harmonic motion .
I endeavoured to find whether X , the wave-length of ripple ( in inches ) was directly proportional to v , the maximum velocity of the water relatively to the bottom ( in inches per minute ) during the oscillatory motion ; also the values of vl and 1883 .
] On the Formation of Ripple-mark . .
v2 , the least and greatest velocities of the water compatible with the formation of ripple-mark .
The following are the results X-i-v .
Feet per Bee .
\#171 ; i\gt ; *0031 ... .
*51 to *56 *0027 to -0028 ____ *46 to *51 .0024 ____ *43 to *49 .0021 ____ *43 to *52 .0023 ____ *50 .002 ____ *56 \#151 ; ... .
more than *42 Mean *00245 min. ... .
*503 ... .
1*2 or *147 second .
It appears therefore that ripples are not formed if the maximum velocity of the water relatively to this particular sand , estimated on the above hypothesis , is less than half a foot or greater than a foot per second ; and that if v be that maximum velocity in inches per minute , the wave-length of ripples generated is .00245\lt ; i ?
" or *147v when v is measured in inches per second .
The results seem as fairly consistent with one another as could be expected .
It will appear from section 2 that the maximum velocity of the water , as estimated on the hypothesis that the water as a whole executes a simple harmonic oscillation relatively to the bottom , does net give the maximum velocity of the water in contact with the sand relatively thereto .
The quantity called v is not in reality the maximum velocity of the water in contact with the bottom relatively thereto , but it is in fact 6 283 times the amplitude multiplied by the frequency .
Thus we cannot conclude that a current of half a foot per second is just sufficient to stir the sand .
In the state of oscillation corresponding to it is probable that part of the water at the bottom is moving with a velocity much greater than half a foot per second relatively to the sand .
The number of the experiments analysed is insufficient for the accurate determination of the law connecting wave-length , amplitude , and frequency ; but this branch of the subject was not pursued further because other observers ; whose work is referred to in S 3 , have made a number of experiments with this object .
It was after making this set of experiments that I hit on what appears to be the key-note of the whole phenomenon .
A series of ripples extending inwards for some distance having been made by oscillation , and the water having come to rest , the bath was turned slowly and nearly uniformly round .
The ripples were then observed to prolong themselves towards the centre ; this shows that a uniform current is competent to prolong existing ripples .
The Feet per see .
va .
22 Prof. G. H. Darwin .
[ Nov. 22 , uniform onrrent flattened the tops of the ripples , but made the lee .
side steeper .
After being exposed to a prolonged current , the ripples were not only not in course of obliteration , but became somewhat more pronounced .
The sand was then smoothed with the edge of a board ; after the exposure of the sand to a current , marks made with the edge , which at first were too faint to be seen , became by a course of development well-defined ripples .
The whole surface became gradually mottled with irregular chains of ripples of which the weather-side was a gradual slope , and the lee-side was steep .
The appearance was strikingly like that of drifted snow .
As it might be conjectured that there would be eddies or vortices on the lee-side , I made some regular ripple-marks by oscillation , and then exposed them to a current .
I shortly observed minute particles lying on the surface of the sand climbing up the lee-slope of the ripples apparently against stream .
This proved conclusively the existence of the suspected vortices .
If when the bath was at rest , a sudden motion was given in one direction , the sand on the lee-side of each ripple was observed to be churned up by a vortex .
By giving a short and sudden motion , I was able to see the direct stream pile up the sand on the weather-side and the vortex on the lee-side .
Fig. 1 shows the effect of a Fig. 1 .
single short jerk .
Two little parallel ridges of sand were formed , namely ( a ) by the direct stream , and ( \amp ; ) on the lee-side by the vortex , a little below the crest of the ripple-mark .
It is thus clear that casual surface inequalities are accentuated by the combined action of the direct stream and of the vortex .
For the purpose of examining the vortices a glass tube was drawn out to a fine point , and fitted at the other end with a short piece of india-rubber tube .
With this a drop of ink could be squirted out at the bottom of the water .
This method was adopted in all subsequent observations , and it proved very valuable .
It may be worth mentioning that common ink , which is heavier than water , was better than aniline dye ; and the addition of some sulphate of iron to the ink improved its action .
A drop of ink was placed in the furrow between two ripples ; as soon as the continuous stream passed , the ink was parted into two portions , one being sucked back apparently against stream up the lee-sidq of the ripple-mark , and the other being carried by the direct 1883 .
] On the Formation of Ripple-mark .
stream towards the crest .
By observing the limits of the transport of the ink , I conclude that the stream lines were as shown in fig. 2 .
Fio .
2 .
These points being settled , it remained to discover how the vortices were arranged which undoubtedly must exist in the oscillatory formation of regular ripples .
The rapidity of the necessary oscillations made this a task of some difficulty .
S2 .
On the Formation of Ripple-mark Oscillation .
The observations were made in two different ways .
In the first of these , which also ultimately proved to be the most successful , the ripple-mark was made in a glass trough about 1 foot long , 5 inches wide , and 6 inches deep .
In order to observe the formation of the ripples absolutely in profile , a sheet of glass fitting the trough was placed to stand on four short corks at the bottom .
The trough was then put to rock on two corks , one at each side , on the line which bisects its length .
Two other slightly shorter corks were put at the ends ; these served as stops , and only allowed it to rock through a very small angle .
The trough was placed on a window-sill with a strong light outside , and was gently rocked by hand .
When the trough is half filled with water , and sand is sprinkled on the glass plate , it is easy to obtain admirable ripple-marks by gently rocking the trough .
When a very small quantity of sand is sprinkled in and the rocking begins , the sand dances backwards and forwards on the bottom , the grains rolling as they go .
Very shortly the sand begins to aggregate into irregular little flocculent masses , the appearance being something like that of curdling milk .
The position of the masses is , I believe , solely determined by the f riction of the sand on the bottom , and as soon as a grain sticks , it thereby increases the friction at that place .
The aggregations gradually become elongated and rearrange themselves .
As soon as the formation is definite enough to make the measurement of the wave-length possible , it is found that the wave-length is about half of what it becomes in the ultimate formation .
; . .
... Prof. G. H. Darwin .
[ Nov. 22 , Some of the elongated patches disappear , and others fuse together and form ridges , the ridges then become straighter , and finally a regular ripple-mark is formed with the wave-length double that in the initial stage .
When a drop of ink is put on the glass without any sand , it simply slides to and fro , with perhaps a faint tendency to curdle , but it cannot be caused to form ripple-mark .
This shows that the initial stage when the sand is beginning to curdle is due simply to friction .
When the ink is put upon a flocculent mass it betrays some kind of dance in the water , but the layer of water disturbed is so thin that it would hardly have been possible to detect the law of the motion from this case alone .
When , however , the nature of that motion , as described below , has been discovered , the same kind of motion may be recognised in the dance of the ink over these flocculent aggregations .
I found in the later experiments that it was advantageous to have a very regular ripple-mark .
I therefore sprinkled sand on the sheet of glass , and , before beginning the rocking , I traced regular furrows in it with the point of my finger .
A few oscillations of the trough soon effaced all signs of the artificial origin , and the ripple crests with the bare glass in the furrows , were absolutely indistinguishable , except by perfect regularity , from those produced naturally .
Most of the observations were , however , made with the natural ripples , and it was only towards the end that I adopted this plan in order to save time and to obtain perfect regularity .
In the rocking trough , the water moves whilst the bottom of the vessel is still , save for the small rocking motion .
A second arrangement was , however , made in which the converse is true .
A sheet of plate glass is caused to oscillate in the bottom of a trough with glass sides .
The oscillator is moved by a connecting rod and crank driven by a small water-motor , the throw of the crank is small , and the rapidity of oscillation can be varied within considerable limits .
When sand is sprinkled on the oscillating sheet of glass , phenomena such as described above are again observed , and good regular ripple-mark is formed in the sand .
Although much was learned from this instrument , still the rocking trough was on the whole more useful .
It appeared to be certain from the first set of experiments that ripple-mark was due to eddies or vortices , and the question remained as to how the vortices were arranged in oscillatory motion .
It required some practice , and many hours of watching , to establish the conclusions ; explained below , indeed the phenomena next described were only detected long after that which follows them in this paper .
If a very gentle oscillation be started , the layer of ink on the crest of a ripple-mark becomes thicker and thinner alternately , swaying backwards and forwards ; then a little tail of ink rises from the 1883.1 On the Formation of Ripple-mark .
crest , and the point of growth oscillates on each side of the crest ; the end of the tail flips backwards and forwards .
Next the end of the tail spreads out laterally on each side , so that a sort of mushroom of ink is formed , with the stalk dancing to and fro .
The height of the mushroom is generally less than a millimetre .
Fia .
3 .
Fig. 3 is the best representation I can make of this appearance , which I shall call an ink mushroom .
The first of these figures gives the extreme of excursion on one side , the second the mean position , and the third the extreme on the other side .
The figures show the state of affairs when the oscillation is very gentle , so that the amplitude of oscillation of the main body of water is small compared with the wave-length of the ripple-mark .
The elongated hollows under the mushroom are the vortices , and the stem is the upward current .
If the ink be thick these spaces are clouded , and the appearance is simply that of an alternate thickening and thinning of the ink on the crest .
When one is familiar with this motion , after examining it carefully with gentle oscillation over ripple-mark of some size , the same kind of dance may , I think , be detected in the stage of ripple manufacture after the sand has curdled into elongated flocculent masses .
The oscillations being still gentle , but not so gentle as at first , streams of ink from the two mushrooms on adjacent crests creep down the two slopes into the furrow between the adjacent ridges , and where they meet a column of ink begins to rise from the part of the water whose mean position is in the centre of the furrow .
The column is wavy , and the appearance is strikingly like that of smoke rising from a fire in still air .
The column ascends to a height of some 5 , 10 , or perhaps 20 times the height of the ripple-marks , according to the violence of the agitation .
It broadens out at the top oh each side and spreads out into a cloud , until the appearance is exactly like pictures of a volcano in violent eruption ; but the broad flat cloud dances to and 26 Prof. G. H. Darwin .
[ Nov. 22 , fro relatively to the ascending column .
The ink continues to spread out laterally and begins tcf fall on each side .
In this stage if the ink is not thick it is often very like a palm-tree , and for the sake of a name I call this appearance an ink tree .
The branches ( as it were ) then fall on each side , and the appearance becomes like that of a beech-tree , or sometimes of an umbrella .
The branches reach the ground , and then creep inwards towards the stem , and the ink , which formed the branches , is sometimes seen ascending again in a wavy stream parallel to the stem .
Perhaps a dozen or twenty oscillations are requisite for making the ink go through the changes from the first growth of the tree .
The descending column of a pair of trees comes down on to the top of the mushroom .
I have occasionally , when the oscillations are allowed to die , seen both tree and mushroom , but the successful manufacture of the tree necessitates an oscillation of sufficient violence to render the observation of the mushroom very difficult .
The alternate thickening and thinning of the ink on the crests seems to render it probable that with moderate oscillation the mushroom vortices are still in existence , or at any rate that alternately one and the other is there .
With violent oscillation , when the stem of the tree is much convoluted , as described below , it cannot be asserted that the mushroom vortices exist , and I am somewhat inclined to believe them to be then evanescent .
Each side of the ink tree is clearly a vortex , and the stem is the dividing line between a pair , along which each vortex contributes its share to the ascending column of fluid .
The vortex in half the tree is clearly in the first place generated by friction of the vortex in its correlated mushroom , and is of course endued with the opposite rotation .
The ascending stem of the tree is a swift current , but over the mushroom the descending current is slow until close to the mushroom , when it is seen to be impelled by pulses .
I was on one occasion fortunate enough to observe a mote in the-water which was floating nearly in the centre of a tree vortex , and counted twelve revolutions which it made before it was caught away from its fortunate position .
If the adjoining crests are of unequal height the stem of the tree is thrown over sideways away from the higher crest ; and indeed it requires care to make the growth quite straight .
The ink in the stem ascends with a series of pulses , and it is clear that there is a pumping action going on which renders the motion of .each vortex somewhat intermittent , the two halves of the tree being pumped alternately . .
The amount of curvature in the stem of the tree depends on the amplitude of the oscillation of the water .
Figs. 4 , 5 , 6 , give fair representations of ink trees .
... .
Fto 5 .
Fig. 4 is the palm-tree stage with gentle oscillation , and figs. 5 and 6 represent the appearance when the amplitude is greater .
Fig. 7 exhibits a tree in which the growth is one-sided on account of inequality between the heights of the bounding crests .
Fig. 7 .
Fig. 8 represents a mushroom and a tree which I have occasionally succeeded in observing simultaneously .
Prof. G. H. Darwin , [ Nov. 22 , Fig. 8 .
The ink is propagated along the convolutions of the stem of the ink tree , but the convolutions are themselves propagated upwards , and each convolution corresponds to one oscillation .
The motion of the ink along the convolutions soon becomes slow , but the convolutions become broader and closer .
Thus the upper part of the tree is often seen to be most delicately shaded by a series of nearly equidistant black lines .
A perfectly normal ink tree , made by a very thin stream of ink , would be like the fig. 9 , in which the whole is formed by a Fig. 9 .
single line ; but it is not possible to represent the extreme closeness of the lines adequately .
In the transition from the mushroom stage to the tree stage it appeared to me that it was very frequent that only half the ink tree was formed .
At any rate I have frequently noted the mushrooms and half the tree vortices lasting during many oscillations , and then the other halves of the trees gradually appeared .
This might , of course , be due to an accidental deficiency of ink in an invisible tree vortex , but I have observed this appearance frequently when there is ink at the stem of the tree , and when there seemed no reason why it should only be carried up in one ascending stream and not in the other .
If the agitation is very gentle the sand on the crests of the ripple-marks is just moved to and fro ; with slightly more amplitude , the dance is larger , and particles or visible objects , such as minute air-bubbles , in the furrows , also dance , but with less amplitude than those on the crests .
When the rocking is gentle the oscillation in the 1883 , ] On the Formation of Ripple-mark .
29 furrow appears to be in a different phase from that on the crest , with more violent rocking I did not observe the difference of phase .
The \#166 ; dance is not a simple harmonic motion like that of the main body of the water relatively to the bottom , but the particles dash from one elongation to the other , pause there , and then dash back again .
As the amplitude further increases the furrows are completely scoured out , and the sand on the crests is dashed to and fro , forming a spray dancing between two .
limits .
With violent agitation this dance must have an amplitude of more than half a wave-length .
If the agitation be allowed to subside the dance subsides , and when the water is still the ripple-mark is left symmetrical on both sides .
With extremely violent oscillation all the water becomes filled with flying dust , and it is no longer possible to see what is happening .
This seems to be the condition when the agitation is too strong for the formation of ripple-mark .
It is probable that the rush of water sweeps away the existing ripple-mark , and there is then no longer anything to produce a systematic arrangement of vortices .
In fig. 10 I have tried to exhibit the dance of the vortices by the succession of figs , i to vii .
When the amplitude of oscillation is the same as when the ripple-mark is generated , the series of changes is of the kind shown .
The figures succeed one another in time , but they do not pretend to such accuracy as to represent the stages at rigorously equal intervals .
The dotted waves show a mean contour of the ripple-mark ; they are introduced to show the displacements of the crests relatively to the mean position .
A perfectly symmetrical ripple-mark does not , however , present a. simple harmonic outline , for the hollows are flat and the crests rather sharp .
The convoluted line is the stem of the ink tree , which forms the \#166 ; dividing line of the two vortices ; the curved arrows show the direction of rotation .
In i the water is at its elongation to the right .
The Crest of the ripple-mark is also at its extreme to the right , and the right hand slopes are steep , whilst the left are gentle .
Here the water is at rest \#166 ; except for the vortices , which both tend to carry sand up to the \#166 ; crest .
In ii the general mass of water is beginning its movement to the left ; it carries with it the upper convolutions of the ink tree , but leaves the root very nearly in the same position as in i. The crest of the ripple-mark is but little displaced .
In iii the crest has begun its displacement , so that although the root of the ink tree has begun moving to the left , it is still over the crest .
The convoluted stem .continues to move to the left with increasing velocity , leaving the root behind it over the crest .
Just before the convolutions reach the position over the middle of Prof. G. H. Darwin , [ Nov. 22 , the furrow , the root leaves its crest and moves with very great speed to the left .
In iv the root is just passing under the convolutions .
The whole system is moving with its maximum velocity , but the root outstrips the stem .
The two slopes of the crest are nearly symmetrical .
In v the root has gained so far on the convolutions as to have again reached a crest .
Fig. 10 .
In vi the convolutions have caught up the root , and the crests are being displaced .
Finally vii is a repetition of i in the opposite direction , and the half oscillation is completed .
If in these figures i to vii we take the wave-length of ripple-mark as unity , the amplitude of oscillation of the main body of water is 2T , 1883 .
] On the Formation of Ripple-mark .
31 that of the crests is '7 , and the breadth of the convolutions of the tree is '3 .
The sum of the amplitude of oscillation of the crest , with the breadth of the convolutions , and the wave-length of ripple-mark is equal to 2 , and this is very nearly equal to the amplitude of oscillation of the water , as it ought to be .
The law which governs the intensity of the vortices must be a matter of inference , since I found the motion too rapid to be sure of anything save that the vortices are driven alternately by pulses , and that the motion was most energetic near the elongations .
In i the right hand vortex of a pair must be at its maximum of intensity , and it seems probable that the left hand vortex has a submaximum in consequence of the friction of the water along the dividing line .
During the return motion from i to vii , the left hand vortex must be increasing in intensity , so that it is at its maximum in vii .
Probably the right hand vortex diminishes in intensity from i to v , and then increases to its sub-maximum in vii .
I am not able to say from observation that the vortices which have been described as giving rise to the ink mushrooms actually exist in this state of oscillation , but if they are there , one of them should be found at the point marked with an asterisk in iv .
The figures tell better than words the mechanism by which the ripple-mark is made and maintained , and the cause of the dance of the crests .
The only difficulty is in stage iv , where the root of the tree is in the state of transference from one crest to the next .
In this stage the vortices would seem to be in the act of degrading the ripple-mark , but they are not then either of them at their maximum of intensity , and the time during which this holds good is exceedingly short compared with the whole semi-period of oscillation .
It seems somewhat likely that small vortices are called into existence at the points marked with asterisks in iv , which serve to protect the ripple-marks from degradation during the transference .
Fig. 11 , i to vii , exhibits the dance of the vortices when the oscillation of the water is considerably less in amplitude than the wavelength of ripple-mark .
Here the crests of the ripple-mark are scarcely sensibly disturbed .
Above the crest is drawn the pair of mushroom vortices , the carved arrows showing the direction of rotation being placed outside of the mushrooms ; but I am not able to satisfy myself that they are both in existence during the whole oscillation .
Fig. 8 above exhibits an appearance which I have sometimes seen , which seems to show that they may both exist together with an ink tree .
We must now draw attention to the manner in which the convolutions are added to the ink tree , and thus show the continuity of this fig. 11 for gentle oscillation with fig. 10 for violent oscillation .
In fig. 10 , in ii , iii , iv , a convolution is added , which is unwrapped Prof. G. H. Darwin .
[ Nov. 22 again in v. It is again formed in vi and vii , and then becomes permanent , and is transmitted up the stem of the tree .
Eig .
11 , In fig. 11 the convolution is added in ii , and then remains permanently part of the tree ; but a partial convolution is added in iii and iv , which is unwrapped again in v , vi , vii .
Thus in violent oscillation the convolution is permanently added just before an elongation , and in gentle oscillation afterwards .
It would be easy to construct a figure for an intermediate amplitude in which the convolution is added just at elongation .
If the oscillation gradually increases , the convolutions are permanently added sooner and sooner , and at the same time the formation of convolutions and subsequent unwrapping assumes more and more prominence .
1883 .
] On the Formation of Ripple-mark .
It must be understood that these figures are drawn from the results of long watching of the process .
My attention was at one time directed to one part of the phenomenon , and at another to a different part , and the amplitudes were constantly varied .
I do not pretend to be able to see all these changes in a single half oscillation , lasting barely half a second .
It may appear that I am incorrect in some parts of the construction , but I would ask any one who repeats the experiments not to condemn me hastily , for the constructions which I have given are the results of frequent trials and errors in the attempt to represent the changes observed .
I have not been able to determine exactly the mode of motion in the initial stages of ripple-making , when the oscillation has large amplitude , but when the ripple-marks are still in what I have described as the curdling stage .
If a current be passed over existing ripple-mark a vortex is established on the lee of each ripple ; if the current be reversed the vortex is on the other side .
Thus intermittent opposite currents will form ripple-mark , but probably without giving it a very regular wave-length .
If the intermittence is rapid , the vortex established on the lee-side , when the current is in one direction , is not annulled when the current is reversed , but it will be carried over the crest of the ripple-mark , and will diminish in intensity , whilst the new vortex with opposite rotation is established .
The study of a very gentle oscillation over existing ripple-mark , by means of a drop of ink placed on the ridge , enables us to observe these vortices ( see fig. 3 ) .
I think it depends on the amplitude of oscillation whether both vortices are always in existence and simply vary in intensity , or whether the vortex due to motion to the right is quite annulled during the motion to the left , and vice versa , .
It may be suspected , therefore , that , in the early stages of ripplemaking , when the amplitude of oscillation is large , vortices are set up in the lee of each aggregation of sand , in the same way as if the current were permanent , and that when the current is reversed these vortices are speedily annulled , and a new set on the other side of the aggregations is established .
When a drop of ink is put on an aggregation , and the oscillation is started , the ink forms a layer of not more than half a millimetre in thickness .
It is easy to see that there takes place some kind of rapid oscillation which is not simply harmonic .
It appears to present all the characters of the motion when gentle oscillation is established over ripple-mark of some height , and therefore it is probable that the motion is of the same kind in both cases .
When the aggregations are more pronounced , small correlated tree vortices are set up .
As above stated , it has seemed to me that frequently only half of each tree vortex is set up at first .
VOL. xxxvi .
D Prof. Gr .
H. Darwin .
[ Nov. 22 , I am disposed to regard this as the transitional state from the mode of oscillation , which produces the octave with small height of ripple-crest , to the fundamental with considerable height .
In gentle oscillation over high ripple-mark the tree vortices are , in the first instance , seen to be started by the mushroom vortices , and the same is probably true of the condition we are considering .
If the suggested view as to the mode of transition be correct , then we must suppose that at first every alternate tree vortex is started by its correlated mushroom vortex .
If there be no tree vortices , or if there be only every alternate one , the vortices can pack twice as close as if the trees are symmetrical ; but the existence of a half tree vortex tends to generate its other half , and this half cannot exist normally unless every alternate ripple-mark is removed .
The degradation of the alternate ripple-mark must arise then from the existence of the second half of the ink tree .
In these early stages the phenomenon is not highly regular , and therefore , besides the smallness of the scale and the rapidity of the motion , we have the difficulty of irregularity to contend with .
Other observers have endeavoured to determine the laws connecting the wave-length in the ultimate formation with the various concomitant circumstances , and I shall leave this subject to the following section , where some account of their work will be given .
We may summarise the results of these observations as follows :\#151 ; The formation of irregular ripple-marks or dunes by a current is due to the vortex which exists on the lee of any superficial inequality of the bottom ; the direct current carries the sand up the weather slope and the vortex up the lee slope .
Thus any existing inequalities are increased , and the surface of sand becomes mottled over with irregular dunes .
The velocity of the water must be greater than one limit and less than another , the limiting velocities being dependent on the average size and density of the particles .
Existing regular ripple-mark is maintained by a current passing over it perpendicular to the ridges .
A slight change in form ensues , the weather slope becoming less steep and the lee slope steeper .
The ridges are also slowly displaced to leeward .
The regular ripple-mark may also thus be somewhat prolonged , so that although a uniform current cannot , as I believe , form regular ripple-mark , yet it may increase the area over which it is to be found .
Regular ripple-mark is formed by water which oscillates relatively to the bottom .
A pair of vortices , or in some cases four vortices , are established in the water ; each set of vortices corresponds to a single ripple-crest and the vortices oscillate about a mean position , changing their shapes and intensities periodically , but not with a simple harmonic motion .
The successive changes in the vortex motion , whilst ripple-mark 1883 .
] On the Formation of Ripple-mark .
is being established , and when the amplitude of oscillation over existing ripple-mark varies , are complex .
As far as I have been able to determine , the following is an account of the phenomena:\#151 ; We begin with variation in amplitude of oscillation over existing regular ripple-mark , where the height of the undulations is not a very small fraction of the wave-length .
When the amplitude of oscillation is small compared with the wavelength , a pair of small vortices are established above the crest of ^ach ripple-mark , rotating in opposite directions .
In the mean position the upward current is over the crest , and the current of water tends to carry up sand from each furrow to the crest .
The dividing line of the vortices oscillates , but the bottom of the line has much less amplitude of oscillation than the top , so that the dividing line is alternately inclined to one and the other side of the vertical .
The vortices are thus carried backwards and forwards over the crest of the ripple , but the current always tends to maintain the crest , merely displacing very slightly the position of the highest point .
The vortex which is on the lee-side is more intense than the other .
We will call these the primary vortices .
( See fig. 3 .
) Suppose now the amplitude of oscillation to be somewhat larger ; then the primary vortices by their friction on the adjacent water generate two other vortices .
The upward current in these secondary vortices has its mean position over the middle of a furrow , and the current comes down immediately over the upward current of the primary pair of vortices .
It appears that sometimes only every alternate one of the secondary vortices is established .
The upward current of the secondary vortices oscillates with a motion which is very far from being harmonic .
It remains at its elongation for a long time and then darts across to the other elongation .
( See fig. 11 .
) During this mode of oscillation the primary vortices are carried much further backwards and forwards over the crests .
With still larger amplitude of oscillation it is no longer possible to distinguish the primary vortices , and the secondary vortices increase in intensity .
It seems probable that the primary vortices are no longer both in existence during the whole oscillation , but that they are alternately created and annulled , so that when one exists the other does not .
If this be so the vortex which exists is that which is on the lee side of the ripple in the state of motion at the instant .
With strong oscillation the secondary vortices apparently do all the work , and the primary vortex , if it exists , only exists for a short time , whilst it may serve as a protecting vortex to the ripple-crest , during the rapid transference of the dividing line of the secondary vortices from one crest to the next .
Each secondary vortex is alternately a vortex under the lee of a ripple-mark as exhibited in fig. 10-Mere description is hardly sufficient to explain the motion .
d 2 36 Prof. ( jr .
H. Darwin .
[ Nov. 22 , .
" With very violent oscillation the ripple-mai'ks are obliterated , and the water is filled with flying dnst .
We now revert to the initiation of ripple-mark .
If the surface be very even , as when sand is sprinkled on glass , when a uniform oscillation of considerable amplitude be established , , the sand is carried backwards and forwards and some of the particles stick in places of greater friction .
As soon as there is any superficial inequality , it is probable that a vortex is set up in the lee of the inequality which tends to establish a dune there .
Such vortices are , , however , too small to be seen .
The return current in the second half of an oscillation maintains the dune , a vortex being established on the-other side , now the lee-side .
As the sand tends to stick by friction in a great number of positions , the sand agglomerates into elongated patches , and the patches are so near to one another that the vortex on either side of one patch just fails to interfere with the next patch .
As the patches elongate and regularise themselves the vortices increase in intensity , and the vortex established on one lee is not obliterated in the return current .
The two vortices are then the primary vortices described above .
As the ripple increases in height by the obliteration of some of the elongated patches , the primary vortices set up the secondary vortices .
Perhaps the normal state of transition is that only one of the secondary vortices is established at first , and that when the other secondary vortex is set up it tends to obliterate every alternate ripple-mark , and thus to generate a ripple of double wave-length .
As the ripples increase in height the secondary vortices become more and more important , and the primary less important .
The final or stationary condition is that described , above as the case of strong oscillation .
It is to be admitted that this history of the successive stages of the formation of ripple-mark is to some extent speculative , but it is the only method of formation which appears to accord with the-various phenomena observed and described above .
It is important to note that when once a fairly regular ripple-mark is established , a wide variability of amplitude in the oscillation is consistent with its maintenance or increase .
No explanation of ripple-making can be deemed satisfactory which does not satisfy this condition .
In this summary no attempt has been made to go over again the various peculiarities of the motion , which have been noted above , such as the dance of the crests and of the convolutions of the dividing line of the vortices .
We must refer the reader back for the consideration of these points .
S 3 .
The WorJc of previous Observers and Discussion .
Some valuable papers on ripple-mark have been lately published1883 .
] On the Formation of Ripple-mark .
The first of these is by Mr. A. R. Hunt.* In it he makes an extensive collection of observations on the natural history of ripple-mark .
As , however , he does not touch at any length on the mode of formation , I have but little to say on his work .
He remarks that regular ripples are due to alternating currents , and that the irregular marks due to currents ought to be distinguished by another name from the regular marks formed by oscillating water .
M. For el , whose paper is referred to below , takes the same view , and describes these irregular marks as dunes .
My own observations seem to accord well with the facts collected by Mr. Hunt .
The second paper is by M. Casimir de Candolle.f His experiments have led him to enounce ( p. 245 ) the following general law : " When a viscous material in contact with a less viscous liquid experiences an oscillatory or intermittent friction , arising from the relative motion of the liquid layer , 1st , the surface of the viscous material is rippled perpendicularly to the direction of motion ; and 2nd , the wave-length is directly proportional to the amplitude of the oscillation .
" The word viscous cannot here have its usual meaning , for sand cannot be called viscous .
The epithet seems to denote that the constituent parts of the material are mobile , and that there is a considerable amount of internal friction .
When oscillations are set up in a vessel containing two fluids of very unequal viscosity , such as tar and water , ripples are formed on the more viscous fluid .
But if the two fluids do not differ widely in viscosity , as mercury and water , water and turpentine , essence of cinnamon and water , ripple-mark is not generated .
If , however , a layer of powder be introduced at the surface of separation , ripple-mark is easily formed .
Ripples were made in sand with a variety of fluids , but with olive oil it was found impossible .
According to the views maintained in the present paper , the viscosity of oil is too great td permit the generation of the ripple-making vortices .
At p. 257 , M. de Candolle writes\#151 ; Chaque ride se termine a la party superieure par un Crete composee des particules les plus legeres .
Tant que dure le balance-ment du liquid , les particules sont animees d'un mouvement pendu-laire qui les transport alternativement de part et d'autre de la crete .
Aussi longtemps que I'amplitude de ce balancement est egale a celle qui a donne naissance aux rides , les particules mobiles parcourent a vo* 34'p I**6 Formation of RiPPle-mark .
" Proc. Roy .
Soc. , April 20 , 1882 , TVn q RlfeS t0\#153 ; teS'l Archives dea Sciences Physiques et Naturelles .
Geneve , No. 3 , vol. ix , 15th March , 1883 .
\gt ; Prof. G. H. Darwin .
[ Nov. 22 , chaque demi-oscillation* toute la distance qui separe l'une de l'antre deux crates cons6cutives .
Ce va-et-vient des particules s'etend jusqu'a une certaine distance au-dessous du somxnet de chaqne Crete , mais son amplitude va en diminuant de haut en bas , en raison du poids plus considerable des particules inferieures .
II en resulte que cbacune de ces cretes mobiles a l'apparence d'une lamelle qui oscille sur le sommet de la ride qu'elle termine et s'etire en meme temps dans le sens du mouvement de l'eau , ce qui donne tout a fait l'apparence d'un corps visqueux , " Lorsque l'amplitude du balancement du liquide diminue , il en est naturellement de meme des excursions de ces lamelles , et si l'on vient a arreter subitement le balancement , les particules composant les cretes mobiles peuvent se deposer entre les rides ou elles forment un systeme de rides secondaires plus minces , intercalees entre celles qui correspondent au maximum d'araplitude du balancement .
" In this passage the dance of the particles is in the first place described as being from one side to the other of the ridge , and this , I believe , is the fact .
This statement is , however , apparently contradicted by what follows , viz. , that the dance is from crest to crest .
I have very rarely seen the intercalated ripple-marks to which M. de Candolle refers , but I venture to think that his explanation is not sound , and that they are formed by the particles of sand which , in violent oscillation , have been caught up by the secondary or tree vortices , carried quite round and dropped at the root of the tree , when the oscillations of the water are dying out .
M. de Candolle arrives at the interesting conclusion that the wavelength of ripple-mark is independent of the nature of the oscillating fluid .
His suggestion that cirrus clouds are ripple-marks between two aerial currents will be referred to below .
The whole paper forms a valuable contribution , and should be read by those who are interested in the subject .
The last paper to which I shall refer is by M. Forel.f He has made extensive observations on ripple-mark , formed both naturally and artificially .
He distinguishes between dunes formed by continuous currents either of air or water and ripple-marks formed by oscillation .
His view accords with the experiments in S 1 above , but he has not apprehended the importance of the vortex in the lee of the * In a letter M. de Candolle tells me that " demi-oscillation " should read " quart d'oscillation ; " but I still do .not see how the ambiguity pointed out below is removed by this correction .
f " Les Hides de Fond .
" Archives des Sciences Physiques et Naturelles .
Gen\amp ; ve , 15 Juillet , 1883 .
M. Forel quotes an article of mine in " Nature " as attributing the formation of ripple-mark to the action of currents in the sea .
My statement was intended merely to imply that shallowness of water is favourable to the formation of ripple-mark .
I had already made a great part of these experiments when that article was written in 1882 .
1883 ] On the Formation of Ripple-mark .
clune , considering that region merely as slack water .
I feel some doubt as to the view that a regular series of dunes may be formed by uniform current ; at any rate , in my experiments the dunes were irregular , and had no definite wave-length .
fTis observations on the circumstances which govern the wavelength of ripple-mark are important .
He finds that the factors which enter are the amplitude and period of oscillation of the water ' ; and a third factor is the maximum velocity of the water , which he takes as identical with the ratio between the two others .
If the water moves as a whole with a simple harmonic oscillation it is undoubtedly true that the ratio of amplitude to period is proportional to the maximum velocity , but the vortices quite disturb this relation .
The maximum velocity of the water relatively to the bottom must depend upon the intensity of the vortices , and this depends upon the height of the ripple-mark .
M. Forel finds that the length and breadth of the vessel have no influence on wave-length , but that it diminishes with increasing depth of water .
This he attributes to a diminution both of the period and of the amplitude of the Oscillation of the water which is in contact with the bottom .
The wave-length increases with the coarseness of the sand .
He remarks that when the ripple-mark is once made , the amplitude of oscillation is without influence on its wave-length .
He draws attention to the two limiting velocities , one too great , and the other too small for the formation of ripples , for which values were found in the experiments of S 1 .
M. Forel explains ripple-mark as the confluence of two dunes , formed alternately by the oscillating currents .
This theory is undoubtedly correct , if somewhat incomplete ; The wave-length , he says , is the amplitude of oscillation of a grain of sand " librement transportee par l'eau .
" This expression requires further explanation ; if it means the amplitude of oscillation of a particle of water at the bottom , when the oscillation is started , and before the ripples have risen , I am disposed to doubt it .
It may mean the distance which an average grain of sand is transported when lying on the surface of other sand , under the like circumstances ; * if so , ripple-marks formed with a thin layer of sand on a sheet of glass should have a longer wave-length than if the sand be thick ; this I am also disposed to doubt .
However this may be , M. Forel considers that the wave-length should vary directly as the amplitude of oscillation , directly as the velocity of the current , inversely as the density of the sand , and inversely as the size of the grains .
Considering with M. Forel that the velocity of the current is * I learn by a letter from M. Forel that this is his meaning .
, Prof. G. H. Darwin .
[ Nov. 22 , the ratio of amplitude to period , we should have the wave-length for any one sand varying as the frequency multiplied by the square of the amplitude .
The few fairly consistent experiments recorded in S 1 , do not accord with this view , for it seemed that wave-length varied as v , which is proportional to frequency multiplied by amplitude .
This I understand to accord with M. de Candolle 's law .
As to the law that the finer the sand the longer the wave-length , M. Forel justly observes that it is in contradiction with the fact that small ripples are formed by fine sand , and large ripples by coarse sand .
But he endeavours to remove the apparent inconsistency by remarking in effect that the larger limiting velocity for fine sand is smaller than the smaller limiting velocity for coarse sand .
There must undoubtedly be truth in this view , but I hesitate to accept it as the whole truth .
Noticing that , in the same sites in the Lake of Geneva , the ripples are always of the same length , he says , " de ces observations il semblerait resulter que l'intensite des vagues a bien pen d'influence sur la largeur des rides ; que la nature du sol est le seul facteur important .
" It appears to me that M. Forel 's view as to the wave-length of ripple-mark cannot be accepted as final , but he has certainly thrown much light on the subject in his interesting paper .
The following considerations bear upon the laws of wavelength :\#151 ; It appeared that in the initial stages of ripple-making , the wavelength is at first only half as long as it becomes ultimately , and that when the layer of sand is thin , the wave-length always remains shorter than if it is thick .
Hence if a little sand is dusted on to the oscillating sheet of glass , it is found that the wave-length of ripple is long in the middle of the patch of sand , and short near the margins .
Thus the patch when ripple-marked presents such an appearance Fig. 12 .
as fig. 12 .
If the sand is thin , this appearance often persists however long the oscillation is maintained .
This shows that wave1883 .
] On the Formation of .
length is a function of the height of the existing undulations ; that is to say , not only of the amplitude of oscillation of the upper part of the vortices , but also of their intensity .
On the parts of the plate where the sand is thick , a continual rearrangement of ripple-mark goes on ; the wave-length extends by the excision of short patches of intercalated ripple-mark , and by general rearrangement .
Finally the sand reaches an ultimate condition as regards wave-length , although rearrangement of ripple-mark still appears to go on for a long time .
Then we find in this final condition most of the sand arranged with a certain fundamental wave-length , but where the sand is thin , patches remain with the octave or half wave-length .
It is not easy to understand precisely the mode in which the oscillation of the water over the undulating bottom gives rise to vortices , but there are familiar instances in which nearly the same kind of fluid motion must occur .
In the mode of boat propulsion called sculling , the sailor places an oar with a flat blade through a rowlock in the stern of the boat , and , keeping the handle high above the rowlock , waves the oar backwards and forwards with an alternate inclination of the blade in one direction and the other .
This action generates a stream of water stem wards .
The manner in which the blade meets the water is closely similar to that in which the slopes of two ripple-marks alternately meet the oscillating water ; the sternward current in one Case , and the upward current in the other are due to similar causes .
We may feel confident that in sculling , a pair of vortices are formed with axes vertical , and that the dividing line between them is sinuous .
The motion of a fish 's tail gives rise to a similar rearward current in almost the same way .
These instances may help u\#163 ; to realise the formation of the ripple-making vortices .
Lord Rayleigh has considered the problem involved in the oscillations of a layer of vortically moving fluid separating two uniform streams.* At the meeting of the British Association at Swansea in 1880 , Sir William Thomson read a paper discussing Lord Rayleigh 's problem.f He showed that , in a certain case in which the analytical solution leads to an infinite value , there are waves in the continuous streams in diametrically opposite phases , and that the vortical stratum consists of a series of oval vortices .
Fig. 13 illustrates this mode of motion .
The uniform current flowing over existing ripple-mark exhibits almost a realisation of this mode of motion , one of the streams of fluid being replaced by the sandy undulations .
The same kind of motion must exist in air when a gust of wind blows a shallow puddle into standing ripples .
* " On the Stability or Instability of certain Fluid Motions .
" Proc. Lond. Math. Soc. ( Feb. 12 , 1880 ) , vol. xi , p. 57 .
t Nature , Nov. 11 , 1880 , pp. 45-G , and see correction on p. 70 .
42 On the Formation of Ripple-mafk .
[ Nov. 22 , Fio .
13 .
It seems probable that what is called a mackerel sky is an evidence of a closely similar mode of motion ; and this agrees with M. de Candolle 's suggestion that cirrus is aerial ripple-mark .
The layer of transition between two currents of fluid is dynamically unstable , but if a series of vortices be interpolated , so as to form friction rollers as it were , it probably becomes stable .
It is likely that in air a mode of motion would be set up by friction , which in frictionless fluid would be stable .
If one of the currents of air be colder than the other , a precipitation of vapour will be caused in the vortices , and their shapes will be rendered evident by clouds.* The direction of striation and velocity of translation of mackerel clouds require consideration according to this theory .
If the velocity of the upper current be u , and of the lower current be \#151 ; u , the interposed vortices have a velocity zero , and have their axes perpendicular to the velocity u. Hence if relatively to the earth the rectangular components of the upper current are u+w , v , and of the lower are \#151 ; u-\-w , v , the component velocities of the vortices are w , v , and their axes are parallel to the component v. Therefore the striations should be parallel to that direction in which the two currents have equal components , and the Component velocity of the clouds parallel to the striations should be equal that of either current in the same direction .
The resultant velocity of the clouds is clearly equal to a half of that of the two currents , and the component velocity of the striations perpendicular to themselves is the mean of the components of the two currents in the same direction .
[ If one of the currents veers the axes of the vortices are shifted .
* I had suggested that the centrifugal force of the vortical rotation must produce rarefaction , fall of temperature , and precipitation of vapour .
I have to thank Professor Stokes for pointing out that the fall of temperature would necessarily be so small as only to cause precipitation if the air were almost completely saturated .
I think the fall of temperature at the centre of the vortex might be between a hundredth and a fiftieth of a degree Centigrade .
[ Sir W. Thomson tells me that he had given this vortical explanation before the B. A. in 1876 .
\gt ; The volume , however , gives no abstract .
He explains the formation of cloud as due to the upward motion in the vortices , and consequent rarefaction.\#151 ; Jan. 4 , 1884 .
J 1883 .
] On the Atomic Weight of Titanium .
The existing clouds will be furrowed obliquely , and the vortical stratum will be cut up into diamond-shaped spaces , determined by the intersection of the old and new vortex-axes .
This would explain the patch-work arrangement commonly observed in mackerel sky .
May not the lengths of the patches give a measure of the rate of veering of one of the currents ?
\#151 ; Jan. 4 , 1884 .
] The above account of the formation of ripple-marks shows it to be due to a complex arrangement of vortices .
The difficulty of observation is considerable , and perhaps some of the conclusions may require modification .
I hope that other experimenters may be induced to examine the question .
Lord Rayleigh has shown me a mathematical paper , as yet unpublished , in which he has considered the formation of aerial vortices over a vibrating plate .
It seems possible that an application might be made of similar modes of approximation to the question of water oscillating over a corrugated bottom .
Even a very rough solution would probably throw much light on the exact changes which the ripple-making vortices undergo , and any guidance from theory would much facilitate observation .
" On the Atomic Weight of Titanium .
" By T. E. THORPE , F.R.S. Received November 7 , 1883 .
( Preliminary communication .
) The sto'ichiometrical quantities which we ordinarily term atomic weights are not only the fundamental constants of chemical calculations ; their relations as mere numbers are of the highest significance in connexion with our conceptions concerning the essential nature of matter .
The recent publications of Becker and Clarke in America , and of Lothar Meyer and Seubert in Germany , have served to demonstrate ou how slight an experimental basis a large number\#151 ; the greater proportion , it must be confessed\#151 ; of the accepted values of these constants really depend .
A notable instance of this fact is seen in the case of titanium .
The atomic weight of this element was determined by Rose in 1829 , and by Pierre in 1847 with the following results :\#151 ; Rose ... ... ... . .
48 T3 and 49 '58 , Pierre ... ... ... 50 '25 .
The commonly accepted value of titanium is that founded upon the experiments of Pierre ; the atomic weight adopted by MendelejefE in the series based upon his periodic law is 48 , a number which finds

