119 Experiments on Stream-line Motion in Curved Pipes .
By John Eustice , B.Sc. , Professor of Engineering , Hartley University College , Southampton .
( Communicated by Sir Joseph Larmor , Sec. R.S. Received January 26 , \#151 ; Read February 16 , 1911 .
) Authors previous Experiments.\#151 ; In the paper read by the author on the " Flow of Water in Curved Pipes , " before the Royal Society on June 2 , 1910 , it was shown that even a small curvature in the length of a cylindrical pipe affected the quantity of flow of water through the pipe .
The effect at velocities below the critical velocity for a straight pipe was most remarkable , inasmuch as the experiments showed that in coiled pipes there was apparently no critical velocity region , whilst in less pronounced curves , where the critical velocity is not entirely absent , a very slight sinuosity of the pipe lessened the flow .
This was shown by the increase in the value of the index n in the formula , s = Krn/ m from n \#151 ; 1 , for perfectly straight pipes , to n = IT to 1*2 for pipes slightly curved or sinuous .
Here or is the hydraulic gradient , v is the velocity of flow , mis the hydraulic mean radius , and K is a constant .
In an attempt to discover the cause of this departure from the law of flow in straight pipes , the author had tried Prof. Osborne Reynolds ' colour-band test in a coiled glass tube , but the arrangements were of a primitive character and the results obtained were not decisive .
At the suggestion of Sir Joseph Larmor the colour tests have been repeated with specially made glass tubes , in which the stream motion could be traced by the introduction of coloured water through capillary nozzles .
Arrangements for Coloured Filaments.\#151 ; The following arrangement was adopted:\#151 ; A brass tube A ( fig. 1 ) of about 2 cm .
internal diameter was fitted internally with a piece of brass B of boat-shaped section , into which six separate supplies of dyed water could be introduced through the nipples shown in fig. 1 .
The dye was supplied from the open tubes T , the height of each of which could be adjusted so as to allow the dye to flow in a gentle stream from the nozzles N , U , which were 1'25 mm. external and 0'3 mm. internal diameter .
The supply of water came from a tank through a cock at C ; to this was connected the pipe which contained the nozzles .
The water from the tank in its passage past the nozzles carried the dyes with it .
The direction of flow is shown by the arrows , the velocity of flow was regulated by a pinch-cock PC placed on a flexible tube at the outlet end of the tube under 120 Prof. J. Eustice .
Experiments on [ Jan. 26 , observation .
Glass tubes of various shapes could be connected at the joints J , J , and the stream-line motion made visible by the dyes could be observed and photographed or sketched .
In older to trace the paths of the streams of dye from each nozzle separately , six different colours were used , the nozzles and supplies of colour being so arranged that the stream from each could be observed separately ; by an adjustment of the tube which contained the nozzles , every part of the tube could be investigated .
In fig. 1 , six streams of dye are shown passing through the conical glass connecting tube JJ ; the paths of two only of the filaments are traced in the U-tube .
In fig. 2 the complete paths of six such colour bands are traced .
The first tubes experimented on were U-shaped , as shown in figs. 1 and 2 ; the experiments in these tubes are described in detail , since they illustrate the methods adopted for other curved tubes .
For the purpose of observing and photographing the stream-lines , the U-tube , with its central plane horizontal , was placed in a tank ; mirrors with their planes at an angle of 45 ' with the plane of the U-tube were arranged in positions for obtaining the side and end elevations .
In order to lessen the effect of refraction , the tube was completely covered with water ; photo Stream-line Motion in Curved Pipes .
1911 .
] graphs were taken of the tube and the reflections , and sketches were made from which fig. 2 has been drawn .
The photographs and drawing show the character of the stream-lines , but it is impossible to reproduce the extremely beautiful effects due to the interlacing of the colour bauds .
Although each band kept distinct from the others it was necessary when tracing out the stream-lines to shut off all the colour bands except the one under observation .
Results obtained inU -tubes.\#151 ; A stream filament a ( figs. 1 and 2 ) if it is in the central plane of the tube and close to the outer wall of the bend , breaks up into two parts immediately it reaches the curve of the bend , each of which leaving the central plane in opposite directions and following the wall of the tube crosses to the inner wall , and coming towards the central plane the divided filament forms the loop shown in the end elevation of the bend ( fig. 2 ) ; the filament is now spread out into a narrow band and passes ' through the outlet limb , the path taken being shown by the lines a , a. Stream filaments which are near the outer wall of the tube and just above ( or below ) the central plane will cross the bend as shown by the line b ( fig. 2 ) , and follow the surface of the tube , keeping above ( or below ) the central plane .
The behaviour of other filaments which are not in the central plane of the tube may be seen from the paths of c and d. Both these lines strike the outer wall of the bend and cross over near the surface of the tube , c to the inner wall of the bend and d to the inner wall of the outlet limb .
The filaments e and / pass around the bend and do not strike the walls .
Several U-shaped tubes , right angled bends , and other tubes bent in curves of large radius were experimented on ; some of these are illustrated in the diagram ( figs. 3 to 7 ) .
The tubes are drawn accurately to scale .
The radius in centimetres of the centre line of the curve is in each case given by the number printed at one end of each tube .
The drawings of the stream-lines in fig. 2 are from sketches and photographs of some of the experiments .
In fig. 3 ( 2 ) the filament a strikes the outer wall at A ( which is near the central plane of the tube bend ) , where it spreads out into a flat band and after following the wall of the tube it strikes the inner wall at A ' and then takes the form of path shown ; a section of such a colour band after leaving A ' is usually curved .
In fig. 3 ( 4 ) the colour bands \amp ; BB ' and eCC ' illustrate the effect of increasing the velocity , cCC ' being at a mean velocity of flow of 3'7 cm .
per sec. , whilst \amp ; BB ' is at a velocity of 11 cm .
per sec. In fig. 3 ( 6 ) the band dDD ' after leaving D ' breaks into branches , the main portion D flowing near the centre of the outlet limb whilst T\gt ; 'd " flows near the inner wall of the tube .
Stream-line Motion Curved Pipes .
1911 .
] In fig. 3 ( 8 ) the line/ FF ' , taken at a velocity of 4'6 cm .
per second , is of the same character as the lines previously described ; after striking at F , it spreads out into a wide band , collects at F# and leaves the tube in a fairly regular band of colour .
The line rEE ' shows the effect of increasing the velocity to 18 cm .
per second .
In fig. 3 ( 10 ) the line gGQ ' is at a velocity of 17'5 cm .
per second and the line / iHH/ is at a velocity of 5'5 cm .
per second , just sufficient for the colour band to strike the outer wall of the curve at H " , the band then spreads as shown in the sketch .
The effect of increasing the velocity in a curved tube is to increase the curvature of the filaments ; for example the band H'H " which is shown touching the tube at H " would pass down the outer straight part of the tube without touching it if the velocity of flow is increased ; other examples of this are given in the right-angled bend fig. 4 ( 2 ) and in fig. 1 , Results in Tubes of large Radius.\#151 ; The same general results are obtained in tubes in which the radius of curvature of the tube length is large ; fig. 3 ( 25 ) illustrates the effect of doubling the velocity of flow in a tube 1 cm .
diameter bent to a curve of 25 cm .
radius ; the line/ J ( velocity 5-5 cm .
per second ) shows the repeated crossing of the colour band ; il is at double the velocity of flow ( 11 cm .
per second ) .
The increased curvature of the band at the higher velocity is again apparent .
The filament h in fig. 3 ( 50 ) and l in fig. 3 ( 100 ) show that , even when the curvature is very small , if the curve is sufficiently long the stream-lines will all strike the outer wall and follow the surface of the tube to the inner wall .
The filament m in fig. 3 ( 100 ) enters the straight part of the tube in the central plane close to the outer wall ; it spreads out into wide bands both above and below the central plane , and after crossing the tube opens out into two bands which remain uncombined for the whole remaining length of the tube , the inner radius of the bend being quite free from colour .
In a longer tube these bands would cross over again to the outer radius , as is shown in the coiled tube , fig. 5 , where all the stream-lines repeatedly cross from the inner wall to the outer wall and back .
In a straight tube the colour bands remain distinct and the surrounding water is not tinted ; in curved tubes after the colour bands strike the walls of the tube , although the bands can be distinctly traced , some of the colouring matter is dispersed into the surrounding water , the outlet water being tinted .
In a coiled tube of several convolutions the distinctive character of the colour filament is gradually lost and after passing through several coils can scarcely be traced .
Results in right-angled Bends.\#151 ; In the series of right-angled bends illustrated Prof. J. Eustice .
Experiments on [ Jan. 26 , in fig. 4 , the lines a , b , c , in ( 2 ) were taken at a velocity of 105 cm .
per second ; when the velocity was increased to 15*2 cm .
per second the positions of all the lines changed slightly , the line a , which at the lower velocity was in the position of the regular curve shown , broke up at the higher velocity into two branches , one of which formed a spiral vortex near the central plane of the tube and interlaced with the line c. The position of the filament b was arranged so that it struck the extreme part of the outer curve , where it scattered as shown ; at the increased velocity the line b passed through the outer straight tube without scattering .
In ( 4 ) the velocity is 7*3 cm .
per sec. , the line / strikes the bend below the central plane and crosses over close to the surface of the tube ; other lines , such as e , which are above the central plane , follow paths shown at eE , whilst the lines represented by d nearer the central plane and the inner surface of the tube pass around the bend in an unbroken curve .
In ( 6 ) the velocity is 12'5 cm .
per sec. , the line g crosses under the other line , which passes through in an unbroken curve .
In ( 8 ) the velocity is 8 cm .
per sec. , the line h is close to the side of the tube near the central plane , it spreads immediately on entering the curved part , crosses the tube , and collects inside the bend , then leaves the bend in separate filaments ; the other stream-line passes freely through the tube .
In ( 10 ) the velocity is 10 cm .
per sec. ; the results are very similar to that shown in ( 4 ) .
The results obtained in sharp right-angled elbows are shown in figs. 6 and 7 .
The curious looping of the lines in fig. 7 will be understood from the somewhat similar loop shown in fig. 6 .
The table on p. 125 is given to indicate the range of velocities employed , but the experiments were of a qualitative rather than of a quantitative character .
The U-tube , fig. 2 , is 1*7 cm .
internal diameter , all the others are 1 cm .
diameter .
The lengths of the curved part and the total lengths of each tube are given in Column 2 .
In each case the straight parts of the tube at both ends are of equal length .
Except in fig. 2 a connecting tube , 20 cm .
long , was used in the position JJ , fig. 1 ; this tube tapered from 2 cm .
to 1 cm .
diameter for 10 cm .
of its length , and was 1 cm .
diameter for the remaining 10 cm .
For fig. 2 the connector was only 4 cm .
long , and tapered from 2 cm .
to 1'75 cm .
diameter .
The head in the tank varied from 20 to 30 cm .
Precautions taken in the Experiments : ( ) To ensure the equal velocities of flow of the colour filaments and the surrounding water.\#151 ; Before using 1911 .
] Stream-line Motion in Curved Pipes .
1 .
Fig. 2 .
Length of tube , cm .
3 .
Radius of curve , cm .
4 .
Temp. 'C .
5 .
Mean velocity of flow in diagram , cm .
per sec. 6 .
Turbulent motion , cm .
per sec. 7 .
Critical velocity by Reynolds ' formula .
Curved part .
Total length .
Fig. 2 7-8 34 -8 2-5 17 -0 3 5 7\#151 ; 8 12 *8 Fig. 3 ( 2 ) ... 6-3 50 -0 2*0 18 -0 a 10-0 r x o .^7 1 17\#151 ; 27 21 5 Fig. 3(4 ) ... 12 -6 50 -0 4-0 18-0 { c 11-0 } 17\#151 ; 21 21 -5 Fig. 3 ( 6 ) ... 18 -8 50 -0 6 -0 18 -5 d 7-8 18\#151 ; 20 21 -3 Fig. 3 ( 8 ) ... 24 -1 50-0 8 -0 18-0 f e 18 *0 \ If 4-6 J 18\#151 ; 23 21 -5 Fig. 3 ( 10 ) ... 31 4 50 *0 10 -o 18-0 J \lt ; 7 17 *5 1 1 h 5-5 J 20\#151 ; 26 21 -5 Fig. 3 ( 25 ) ... 50-0 60-0 25 0 19-0 fi 11-0 \ \j 5-5 / 19\#151 ; 21 21 -0 Fig. 3 ( 50 ) ... 50 -0 60 -0 50 -0 18 0 k 10-0 20\#151 ; 21 21 -5 Fig. 3(100 ) ... 50 -0 60-0 100 -o 18 -0 J l 12-6 \ \ m 6 '4 J 20\#151 ; 21 21 *5 Straight tube \#151 ; 60-0 CO 16 -0 \#151 ; 19\#151 ; 21 22 -5 Fig. 4(2 ) ... 3-1 13 -1 2-0 19 -0 10 -5\#151 ; 15 -2 18\#151 ; 20 22-5 Fig. 4(4 ) ... 6-3 16 -3 4-0 19 -0 7-3 19\#151 ; 21 22 -5 Fig. 4 ( 6 ) ... 9-4 19 -4 6-0 19 -0 12-5 18\#151 ; 20 22 5 Fig. 4(8 ) ... 12 -6 22 -6 8-0 19 -0 8-0 19\#151 ; 21 22 -5 Fig. 4 ( 10 ) ... 15 7 25 7 10 -o 19 -0 10 -o 19\#151 ; 21 22 5 the arrangement of nozzles and colour supply sketched in fig. 1 , the author designed an apparatus for a single filament of colour which acted automatically so as to give equal pressures on the colouring liquid and the flowing water ; it could be so arranged that at any pressure the coloured liquid would ooze very gently into the flowing stream .
In other tests the colour was allowed to accumulate in the tube BC , fig. 1 , when no water was flowing ; the colour was then shut off and water allowed to flow through the tube , the paths of the stream-lines could be distinctly traced , following the same general directions as in the case of the regular filaments .
The motion of the solid particles referred to on p. 129 is an additional proof of the effect produced not being due to the filaments having a higher velocity than the surrounding water .
The arrangement described above for single filaments was used for high pressure experiments in metal pipes , which were 150 cm .
long and 1 cm .
diameter .
Between these pipes and the colour distributing nozzles was interposed a converging tube which was connected to the metal pipe by a glass tube , 15 cm .
long , of the same diameter as the metal pipe ; a similar glass tube was connected to the outlet end of the metal pipe .
When the metal pipe was kept straight the colour filament showed distinctly in the 126 Prof. J. Eustice .
Experiments on [ Jan. 26 , glass tube at the outlet end ; when the metal pipe was bent into a slightly curved form the colour filament no longer appeared in the glass tube at the outlet end as a thread of colour , but , if visible , it was as a flat band of colour , or it was more or less dispersed into the surrounding water .
Long , straight , glass tubes of from 1 to 2 metres long were substituted for the metal pipe , and when curved tubes were connected at the outlet end the same effect was produced as when the intervening connection was only a few centimetres long .
( b ) Establishment of a steady regime.\#151 ; From the preceding section it will be seen that in some of the experiments the tubes were sufficiently long to ensure that the filaments and water had settled down to a steady In all the experiments the brass tube A , fig. 1 , containing the nozzles for colour filaments was close to the tank , but between A and the tube experimented on there was interposed a converging tube , the dimensions of which are given on p. 124 .
This tube was of sufficient length to establish steady flow , as will be seen by comparing Columns 6 and 7 of the table .
Column ( 6 ) gives the commencement of the turbulent flow , Column ( 7 ) the critical velocity as calculated by Reynolds ' formula .
The same general results were obtained at low velocities when a steady regime is produced in a fairly short pipe at velocities far below the critical .
In the U-tube , fig. 2 , the connecting tube was short , and unsteady motion commenced at a much lower velocity than is given by Reynolds ' formula .
The turbulent motion given in Column ( 6 ) and the calculations in Column ( 7 ) are for the straight part of the pipe .
Comparison between the Flow of Water in Bends of Open Channels and of Closed Pipes.\#151 ; Prof. James Thomson , * in his theory of the transfer of solid particles along the bed of a river from the outer to the inner bank , has shown that the centrifugal force of the more rapidly moving water near the surface overcomes that of the water close to the bottom , which is retarded by the friction of the bed of the river .
Prof. 0 .
Reynoldsf has adopted this theory in his work " On Certain Laws relating to Rivers and Estuaries .
" Thomson 's experiments were carried out in open channels , and he has suggested that the theory of flow in curved channels is applicable to pipe bends .
In introducing the theory he states that " a stream flowing along a straight channel , and thence into a curve , must flow with a diminished velocity along the outer bank and an increased velocity along the inner bank .
" The author 's experiments show that the motion at the commence* \#163 ; Roy .
Soc. Proc. , ' May 4 , 1876 , vol. 25 , and June 21 , 1877 , vol. 26 .
t 'Report Brit. Assoc. , ' 1887 , and 'Sci .
Papers , ' vol. 2 , p. 329 .
Stream-line Motion in Curved Pipes .
1911 .
] meant of the curve of a pipe bend is not that of a " free " vortex , for the velocity is greater near the outside than it is near the inside of a curve .
There is another difference between the two cases .
In an open channel there is freedom in a vertical direction , and , as Thomson has pointed out , the difference of pressure between the inner and the outer banks causes the surface of the water to be inclined from the inner to the outer bank of the curve .
No such freedom is possible in a pipe which is running full , hence it is probable that the curvature of the stream-lines is greater and the effect of pressure on the transfer of water from the outer to the inner curve is more pronounced in a pipe than in a channel .
In experiments on U-shaped open channels of semicircular section , about 1 cm .
radius , the author has noticed that the flow of water near the surface is retarded , for if a filament enters the bend just below the surface MN in the position of 2 , fig. 9 , it divides so that a part goes under the stream , as in fig. 9 ; another part of the filament follows a somewhat similar course near the surface of the water and flows towards the centre of the stream , where it is carried around near the central curve of the bend .
Filaments entering close to 1 behave in the same way , except that some of the colour is in this case carried slowly around the bend quite close to the outer radius .
With these differences between open channels and pipes the following explanation may be looked upon as an experimental confirmation of Prof. J. Thomson 's theory w'hen applied to curved pipes .
AB D .
_h .
J Motion in a Plane Sheet Perpendicular to Central Plane of Bend.\#151 ; Consider a set of stream-lines flowing through the straight part of a pipe in a plane sheet which is at right angles to the central plane of the U-tube as shown Prof. J. Eustice .
Experiments on [ Jan. 26 , at B and at b"W ( fig. 8 ) where the circles at I , II , III , and IY represent sections of the tube taken along the radial lines 01 , Oil , OIII , and OIY respectively .
Since the stream-lines near the centre of the sheet are flowing at a higher velocity than those nearer V and it follows that when the sheet enters the bend , the force required to make the stream-lines in the centre of the sheet ( at b in section I ) tend to follow the curve of the bend will be greater than the force required nearer V and b " , that is to say , the pressure between* b and the outer wall of the bend will be greater than the pressure nearer b ' and b " , and a flow of water will take place along the path of least resistance , i.e. from the region of greater to the region of less pressure , in the directions shown by the arrows in II , III , and IY .
The centre portion of the sheet will gradually approach the outer wall , whilst some portion of the water near b ' and b " will be carried round the inner wall of the tube .
Motion in a Sheet in the Central Plane of the Bend.\#151 ; Consider the streamlines flowing in the central plane MN of the tube ( fig. 8 ) .
Changes in pressure above and below the central plane will have no effect in disturbing the position of the lines , since the tube is symmetrical on both sides of the plane MN .
In any other sheet M'N ' which was parallel to the central plane before it entered the curved part , since there are less stream-lines , and the velocities are less than in the plane MN , the pressure at M ' will be less than the pressure at M , and the pressure at NT ' will be greater than the pressure at NT ; hence there will be a flow of water from M to M ' , that is to say , the water close to the surface of the tube , which in a straight tube is at rest , will even in a slightly bent tube be set in motion .
As the flow of water from between a stream-line and the tube wall takes place , the stream-line will gradually approach the outer wall of the bend , the displaced water flowing around the walls of the tube from the outer to the inner wall of the bend .
After entering the bend the sheet M'N ' becomes inclined in the direction of a normal to the curves ( fig. 8 ) , as is shown by the deviation of the filaments from the central plane in the left-hand view of fig. 2 .
Motion of the Filaments near the Walls of the Tube.\#151 ; The order in which the stream-lines flow around the walls is shown in fig. 9 .
A filament which approaches the curve at the outer radius of the bend strikes the tube and divides into two parts , each of which crosses a half circumference of the tube to the inside , as shown at 1 .
A filament at 2 which approaches farther from the central plane MN flows inside the filament 1 .
A filament 3 flows inside 2 , and so on ; this is represented in the lower section of the tube in fig. 9 .
Stream-line Motion in Curved Pipes .
1911 .
] It was often noticed that the colour filaments in the position of 2 and 3 crossed the tube in small waves or ripples , showing the approach of unstable motion , which ultimately leads to the mixing of the colour filaments with the surrounding water .
After reaching the inside of the tube the re-crossing always occurred in the half of the tube in which the filament entered , or in a tube of regular bore , the stream-lines are reflected back so as not to cross the central plane MIST .
This reflection is shown in some of the diagrams as in fig. 3 ( 10 ) and ( 25 ) , and in the coiled tube fig. 5 .
Turbulent Motion.\#151 ; The experiments described above were carried out at velocities sufficiently low to give unbroken motion , but in all cases the effect of increasing the velocity was observed , and , although it was naturally impossible to trace the colour filaments for turbulent motion in the same way as has been done for steady motion , by increasing the quantity of dye or by placing finely divided solid matter in the tubes and carefully increasing the velocity until turbulent motion was reached , it was observed that the same general features hold good for turbulent motion .
For example a layer of sand placed along the inside of the tube at aa ' in fig. 1 , in which the central plane of the tube is vertical , will not go around the outer radius of the U-tube but will follow the path a'a " .
The author has*shown that , when water is flowing at high velocity in a metal pipe bend , if a tube communicates from a ' to a " outside the tube a continuous stream of water is carried through the external tube from a !
to Summary and Deductions.\#151 ; The experiments show that when water is flowing at low velocities in a straight tube of uniform bore , filaments of colour maintain their form and relative positions , but when entering upon a curved portion of a tube some of the filaments spread out into bands of colour and cross to the inner part of the tube , travelling round its section close to the walls .
If the curvature is sufficiently large or the curve sufficiently long , all the filaments would be affected .
Other experiments in U-tubes seem to suggest that slipping may take place concomitant with the surface flow at the walls of the curved part of the tube .
A quantity of dye having been spread over the whole of the surface of the tube , when water was allowed to flow through the tube , even at extremely low velocities , the colour disappeared immediately from the outer wall of the bend , more slowly from the inner wall , and very slowly indeed from the straight part of the tube .
In the experiments it was observed that as the velocity of flow increased there was an increase in curvature of the stream-lines .
It is inferred that the resistance to flow along the tube walls from the outer to the inner wall does not vary as the velocity directly , as in the ordinary viscosity equation , VOL. LXXXV.\#151 ; A. K 130 Experiments on Stream-line Motion in Carved Pipes .
but it varies as the velocity raised to a power n , where n is greater than 1 , and probably greater than 2 .
Except for the stream-lines near the walls of the tube the flow in a curved tube at low velocities appears to be in accordance with the viscosity law .
The increased resistance referred to above applies to the water which has impinged upon the walls and is in its circuit along the walls of the pipe .
If a pipe is not perfectly straight , the flow for the whole section of the pipe cannot be wholly viscous flow , since associated with viscous flow there is always some surface flow , that is to say , there is " skin friction , " the effect of which is to increase the resistance of the pipe and to lessen the total discharge ; or , for a bent pipe , n is greater than unity in the formula S = ~K .
un / m. In the author 's previous experiments referred to on p. 119 , it was shown that the increased resistance due to curvature of a pipe was relatively greater at velocities below the critical velocity than at velocities which give turbulent motion in a straight pipe .
This relatively greater resistance at low velocities is probably due to the generation of " skin friction , " which would not exist in a straight pipe of the same area at the same velocity , whereas the " skin friction " already existing is merely augmented at velocities which would cause turbulent motion in a straight pipe .
Although surface flow and viscous flow may exist together in a curved pipe , it was observed that in U-tubes and in angle pipes whilst the filaments in the straight pipe before reaching the walls of the bend continued unbroken up to the usual critical velocity in a straight pipe , the filaments flowing through the straight pipe after leaving the bend usually broke up at velocities much below the critical velocity .
The surface flow which commenced in the bend generated vortices which persisted in the water flowing through the outlet straight part of the pipe .
This shows that the effect of a pipe bend or of an angle is not only to increase the resistance to flow in the bend itself , but also to increase the resistance in the contiguous straight pipe after the water has left the bend .
This agrees with the results obtained by investigators who have experimented on the loss of pressure in the pipe bends used by engineers .
The foregoing experiments indicate that , where it is beneficial to break up the regular lines of flow in a pipe , a curved pipe is more effective than a straight one .
They show , for example , that in a tubular boiler or in a condenser of a steam engine , curved pipes would be more efficient than straight pipes , for when water is flowing at low velocity in a straight pipe , the water near the centre of the pipe section does not approach the sides during its passage through the pipe , but in a curved pipe the water is continually Secondary y-Rays produced by / 3-Rays .
131 changing its position with respect to the sides of the pipe , and the water which is flowing near the centre at one part approaches the sides as it moves through the pipe , and flowing near the sides it exerts a " scouring " action on the pipe walls , thus increasing the effectiveness of the pipe surface in transferring heat .
Secondary y-Rays produced By J. A. Gray , B.Sc. , 1851 Exhibition Scholar , University of Melbourne , ( Communicated by Prof. E. Rutherford , F.R.S. Received February 4 , \#151 ; Read February 23 , 1911 .
) When the cathode rays of a vacuum tube impinge on any material they produce the X-rays , which are not deviated by a magnetic field , and are much more penetrating than the cathode rays which produce them .
We might , therefore , expect that when the / 3-rays from radioactive substances impinge on a plate , similar penetrating rays would be emitted from the-plate .
Such a penetrating type of rays , the 7-rays , is almost invariably associated with the / 3-rays , but it has generally been thought that these-7-rays are due to the expulsion of the / 8-ray from the radioactive atom .
In-some cases they are certainly not due to the impact of / 3-rays on external !
objects , the 7-rays of radium C being an instance of this .
Here the 7-rays come ; from the radioactive atoms , and in such amount that they effectually mask the possible production of 7-rays by / 3-rays as the experiments of H. Stark * show .
Stark attempted to find whether / 8-rays did produce 7-rays.f He used for this purpose 6 milligrammes of radium bromide contained in a very thin glass tube , which let most of the / 3-rays out .
The 7-rays from this ionised the air in an electroscope , the walls of which were thick enough to absorb all the / 8-rays .
He looked for an increase in the ionisation when various materials were placed just behind the radium .
He found practically no difference in the reading , and , from that and a similar experiment in which he deflected the / 3-rays away from the electroscope by a magnetic field , concluded that no measurable 7-radiation was caused by the ^-rays of radium C. * H. Stark , ' Le Radium , ' February , 1908 , p. 35 .
t In this paper a distinction is drawn between primary y-rays and secondary y-rays .
By primary y-rays are meant y-rays coming from the radioactive atom .
By secondary y-rays , y-rays produced by the impact of / 3-rays on external materials .

