The issuing fluid stream was protected from draughts by the glass bell-jar .sx This was exhausted at approximately the same rate as it filled so that the cloud was removed continually .sx Unless this is done the results will not be comparable with actual gas flow because the density of the smoke cloud is much greater than the density of air or gas as supplied to flues and furnaces .sx If the upper end of the bell-jar is attached to an .sx exhauster , a small negative pressure can be maintained within , and the effects of density .sx to some extent , annulled .sx Table I gives the first results obtained .sx The velocity of air leaving the inner tube was constant , variations being made in the position only of the tube with respect to the outer one .sx The figures in the last column show that a neutral condition exists at some point occupying a position intermediate between the nozzle wall and its axis .sx It is possible to map out the shape of the air or gas stream issuing from the inner tube ; the results obtained when the tube was situated 2.05 and 2.50 in .sx below the mouth of the nozzle are shown in the scale drawings , Figs .sx 3A and B. It will also be noted that , although the pressure along the axis falls progressively after leaving the mouth of the small tube , the pressure along the inner wall passes through a maximum , except in those cases where L is small .sx This fact is important , being closely allied to the question of eddying in the immediate vicinity of the exit of the small tube , especially when the lower end of the large tube is partly or completely open to the air .sx In the latter case the arrangement is a small-scale representation of a furnace port in which primary air mixes with the gas by entrainment .sx Considering the cross-section of the gas or air stream at any point , it will be apparent that this is not a maximum in the plane of the nozzle mouth , being greater at some point lower down .sx ( In the case of very low velocities giving pure streamline flow this is not correct but , as such conditions do not often arise in industrial practice , they may , for the most part , be neglected .sx ) The reason for this narrowing of the fluid stream at its moment of exit is to be sought in entrainment of outside air drawn in around the nozzle mouth .sx The velocity at exit is substantially greater than that immediately below , although it is not easy to show this without delicate measuring apparatus .sx To make further investigations a Pitot tube bent at right angles and arranged as in Fig. 1B was employed .sx Such a tube will record the sum of both static and dynamic pressures , whereas the tube previously described measured static pressures only .sx In these experiments .sx the pressure of supply and consequently the air velocity was varied progressively , and measurements were taken of the pressure at different points over the cross-section of the tube in the plane of its mouth .sx The recordings are given in Table II .sx The plus signs in the table indicate readings greater than the maximum value which the manometer would register .sx These results indicate clearly how , as the velocity is reduced , so also is the cross-section of the gas stream .sx due to a lessening of turbulence .sx On moving the manometer towards the centre line , the negative pressure existing close to the nozzle wall falls rapidly until a neutral point is reached , when it changes sign .sx Considering now the case of a nozzle in which the distance L is great compared with the internal diameter ( greater than 12 approximately ) , the high velocity of the fluid stream issuing from the inner tube is soon replaced by a comparatively low velocity in the large tube which remains fairly constant .sx In this case the distribution of the fluid stream velocity over the nozzle cross-section will be elliptical or parabolic .sx The velocity distribution will depend very largely , however , on the physical characteristics of the nozzle wall .sx If this is smooth inside , then it has been shown by Stanton that , up to a radius equal to 0.8 times that of the nozzle , the curve of distribution is a parabola .sx If the walls are very rough , the friction is proportional to the square of the velocity and the curve is parabolic right up to the full nozzle radius .sx Furnace ports and nozzles , being made of refractory material , are usually of a rough nature and must be classed somewhere between the two extremes mentioned above .sx From a number of measurements the author is satisfied that sufficiently accurate results for practical purposes in the case of medium rough pipes and ports are obtained if the velocity is assumed to have an elliptical distribution .sx Referring to Fig. 4 , if V is the velocity along the inside of the walls and V + v that along the axis , it is required to know the position in which the mean velocity U is located .sx Taking a point x distant r from the axis , then if the radius of the nozzle is R velocity at , also , .sx The mean ordinate of an ellipsoid is , therefore the mean velocity and , so that r = 0.745R. The position of the mean ordinate is , therefore , fixed with respect to the axis and has the above constant value .sx The quantity of fluid flowing is then given by ( mean velocity ) ( area ) = UA = .sx In the case of a long pipe having parallel sides , it is fairly safe to assume that the fluid stream will completely , or almost completely , fill the whole of the cross-section .sx In considering a nozzle where there is any constriction at the exit , however , care must be exercised because the velocity at discharge may be greater , or rather the quantity of fluid discharged per unit time may be less .sx than would be anticipated .sx This is due to the effect of the vena contracta , and is of the utmost importance when considering high velocities .sx With low velocities it is probable that most of the cross-section of an orifice will be filled , even if of a very bad design from the point of view of efficiency of discharge .sx With high velocities the reduction in the discharge may be very great .sx The usual formula for the quantity of fluid passing through an orifice in unit time is , where w is area of orifice , theoretical velocity of flow , a coefficient of velocity , and b coefficient of .sx contracted vein .sx The coefficient a is the ratio between the actual velocity and the theoretical velocity , and is usually from 0.95 to 1.00. The other coefficient b is the ratio between the area of the fluid stream at the moment of discharge and the area of the orifice w. Its value is very variable , depending on the shape of the orifice , and may attain a minimum of 0.64. .sx Determination of the point of critical velocity .sx By progressively raising the velocity , it was found that , with low velocities and streamline flow , the fluid stream was nearly parallel sided for some distance ; after a few inches the stream became erratic due to the effect of external eddies and currents , and was soon obliterated .sx On raising the velocity , little change occurred in the appearance of the stream , except an increase in the length of the parallel part .sx A point was soon reached when the flow became turbulent and the fluid stream was then of the form previously described .sx Experiments were also performed using a nozzle similar to that used in the other part of this work , consisting of a large tube fed through a smaller one .sx The results obtained with such a nozzle are given in Table III , and indicate that only at comparatively low velocities is the flow streamline .sx Such velocities are outside those met with in industrial practice ( excluding some cases of draught velocity) .sx It will be noted that the lower the inner tube is situated within the larger , the lower is the critical velocity .sx This is interesting , and is probably caused by the slight negative pressure existing in the region of the orifice of the small tube , due not to entrainment ( which is absent in stream-line flow ) but to actual friction between the moving and stationary fluids .sx If the results are plotted they will be found to lie on a smooth curve .sx Provided the shape of the tube carrying the fluid stream is kept constant , the diameter appears to have little effect on the critical point .sx A number of experiments were made in which the smoke was ejected from simple nozzles of varying shape The results obtained with turbulent flow were similar to those given by water flowing from a jet , and showed the great advantage of the parallel-sided or slightly convergent form over any type consisting of a simple aperture in a plate .sx The convergent form gave the highest coefficient of discharge .sx Summarising these results , it becomes apparent that , except when the flow is streamline , the fluid stream is seen to increase in diameter almost immediately on leaving the orifice , the increase being due to turbulence and air entrainment , and being dependent on both velocity and the ratio of the diameters of the two tubes .sx In the case of a simple jet emerging into air , the effects produced are somewhat different .sx There is then no restriction to the amount of external air supply available for entrainment , and , in general , there is a greater disturbance around the nozzle , the mixed stream soon assuming a constant and regular configuration .sx In turbulent flow , increasing the velocity increases the angle of the emergent cone .sx From careful measurement it was found that for velocities of about 10 ft .sx per sec .sx the angle was 15 , increasing up to 26 for velocities of 50 ft .sx per sec .sx High velocities have , therefore , the greatest spreading effect on the fluid stream .sx It does not follow , however , that the same will hold good in the case of the ignited gas stream .sx Unless the velocity is very high , the shape of the flame will be governed largely by its velocity of propagation .sx The actual flame surface will obviously remain stationary , and the velocity of propagation and port velocity must be in equilibrium .sx With very high velocities it is possible to blow the flame away from the port ; low velocities , on the other hand , cannot be used with burners supplied with primary air , as there will be a tendency for combustion to proceed inside the burner , and possibly to burn or melt it .sx Inferences drawn from the experiments .sx In any burner of a type similar to that described , it is clear that considerable disturbances occur inside .sx particularly in the neighbourhood of the small tube exit .sx If , therefore , it is necessary to premix some air with the gas , the air should preferably be admitted to the burner through side apertures where the internal disturbance is a maximum ( as is done in the Bunsen burner) .sx Only by such means is it possible to entrain the maximum amount .sx of air at atmospheric pressure for any given gas velocity .sx In such a burner , the mixed fluid stream will nearly fill the whole cross-section at exit , although there is generally a slight narrowing here .sx If now the effect of premixing on the type of flame is examined , it is found that the greater is the primary air/gas ratio , the shorter is the flame , or more particularly the inner cone where most of the combustion is taking place .sx Fig. 5 gives some of the author's results , and Fig. 6 shows curves giving the effect of port velocity on cone height .sx It will be seen that increasing the port velocity tends to decrease the cone height , the latter soon reaching an approximately constant value .sx This is very important because the influence of port .sx velocity , with premixed air , is quite different when secondary combustion air only is used .sx When using a burner with a high primary air/gas ratio , high velocities , up to 100 ft .sx per sec .sx , are sometimes used to give economical working and high local temperatures .sx