GB2078977A

GB2078977A - Device for measuring the density of fluids

Info

Publication number: GB2078977A
Application number: GB8113726A
Authority: GB
Original assignee: Individual
Current assignee: Individual
Priority date: 1980-05-02
Filing date: 1981-05-05
Publication date: 1982-01-13
Also published as: GB2078977B; DE3117279A1

Abstract

This invention relates to devices for measuring the density of fluid and is particularly concerned with providing accurate measurements over large density ranges. Two different forms of device are described. In the first oppositely acting float elements 3, 4 are pivotally interconnected. Element 3 bears a scale and element 4 carries a pointer 9. The buoyancy characteristics of the elements are such as to provide good linearity over the working range; rotation of one element predominating at one end of the scale and the rotation of the other element predominating at the other end of the scale. The second device is in the form of a moulded plastics hydrometer, in which different portions 58, 59 are made of materials having different thermal coefficients of expansions to provide linearity over the working density range. <IMAGE>

Description

SPECIFICATION Device for measuring the density of fluids This invention relates to devices for measuring the density of fluids.

Many devices are known for measuring the density of fluid but they suffer from restricted ranges, non-linearity and/or temperature errors.

It is an object of this invention to provide an improved measuring device, which reduces or removes at least one of these limitations.

From one aspect the invention consists in a device for measuring the density of a fluid having a density within a predetermined range and a coefficient of thermal expansion that varies over this range, comprising a body formed such that when floated in the fluid it gives an indication of the fluid density and having an effective coefficient of expansion which varies over the predetermined range in a sense to compensate for variations in the coefficient of thermal expansion of the fluid.

From a second aspect the invention consists in a device for measuring the density of a fluid having a density within a predetermined range, comprising a body formed such that when floated in the fluid it gives an indication of the fluid density, the body comprising two parts, each having respective buoyancy characteristics, the arrangement being such that the buoyancy characteristic of the first of the parts predominates at the lower end of the predetermined range and the buoyancy characteristic of the second part predominates at the upper end of the predetermined range.

From a third aspect the invention consists in a device for measuring the density of a fluid having a density within a predetermined range, comprising a housing for containing the fluid, a first float element pivotally attached to a horizontal axis in said housing, a second float element pivotally attached to said axis, each of said elements including at least one buoyant volume that is less dense relatively to the elements considered as a whole and has a centre of buoyancy displaced from said axis, said buoyant volume having a density which is less than the fluid density to be measured and at least one weight volume that is more dense relative to the element considered as a whole and has its centre of gravity displaced from said axis and a density greater than the fluid density to be measured, each of said elements being formed so that the radius joining said centre of buoyancy to said axis is angularly displaced from and above the radius joining said centre of gravity to said axis, whereby because of differing buoyancy in fluids of different densities the relative angular positions of said elements about said axis indicate the density of the fluid in which they are immersed and said elements rotate in opposite directions about said axis with changes in the density of the fluids, said first float element being constructed and arranged to respond to equal increments of fluid density within said predetermined range with relatively larger angular increments of rotation about said axis in a lower section of said density range, said second float element being constructed and arranged to respond to equal increments of fluid density within said predetermined range with relatively larger angular increments of rotation about said axis in a higher section of said density range, so that the relative angular displacement of said elements in said lower section of said density range, is determined to a greater extent by said first element than by said second element and, in said higher section of said density range, to a greater extent by said second element than by said second element than by said first element.

From a fourth aspect the invention consists in a hydrometer for measuring the density of a fluid having a density within a predetermined range and a coefficient of thermal expansion that varies within this range, comprising a lower body volume Vx which is equal to the total mass of the hydrometer divided by the density of said fluid at the highest part of said range, and an intermediate scale volume Vy which together with said volume Vx is equal to the total mass of the hydrometer divided by the density of said fluid at the lowest part of said range, and an upper stem volume V2 whereby said volume Vx is comprised of one or more materials so proportioned that the coefficient of thermal expansion fo said volume Vx is approximately equal to that of the fluid at the highest part of said range, and said volume Vy is comprised of one or more materials so proportioned that the coefficient of thermal expansion of the combined said volumes Vx and Vy is approximately equal to that of said fluid at the lowest part of said range.

The invention may be performed in a number of ways, specific embodiments of which will now be described, by way of example, with reference to the accompanying drawings, in which: Figure 1 is a diagram of a weighted float; Figure 2 shows a scale associated with the device of Figure 1; Figure 3 shows an example of individual float angles corresponding to equal increments of fluid density; Figure 4 shows an embodiment of float elements and a resulting density scale; Figures 5and 6showthe embodiment of Figure 4 in a greater detail; Figure 7 shows idealized thermal expansion curves which are typical of those encountered in practice; Figure 8 shows the resultant temperature error in a device of known form; Figure 9 is analogous to Figure 7, but applicable to the present invention;; Figures 10a and lOb show a resultant temperature error in one form of the present invention; Figure 11 shows an alternative to Figure 9; Figures 12a and 12b show the resultant temperature error in another form of the present invention; Figure 13 shows a second alternative to Figure 9; Figures 14a and 14b show a resultant temperature error in an ideal form of the present invention; Figure 15a shows an embodiment of a float element analogous to Figure 6, but utilizing a low-density weight element; Figures 15b and 15c show embodiments of a float element analogous to Figure 15a, but with modified ways of utilizing a low-density weight element; Figure 16 shows a resultant temperature error in an application involving real materials;; Figure 17 shows a portable density measuring device; Figure 18 shows idealized thermal-expansion curves typical of those encountered in practice; Figure 19 shows a stem hydrometer divided into parts significant to the present invention; Figure 20 is analogous to Figure 18, but illustrative of the concept of the present invention; Figures 2 la and 2 ib show an embodiment of the present invention; Figure 22 shows thermal-expansion curves corresponding to the embodiment of Figures 21 a and 21 b; Figure 23 shows an embodiment of the prior art in an application involving real materials; Figure 24 shows an embodiment of the present invention corresponding to Figure 23; Figure 25 shows the actual temperature error of the embodiments of Figures 23 and 24;; Figure 26 is analogous to Figures 18 and 20, but applicable to a different fluid; and Figure 27 shows another embodiment of the present invention appropriate to the fluid of Figure 26.

The use of pivotally mounted, weighted floats for measuring fluid density is known. However, since an understanding of their principle of operation may be required by the specification hereinafter, it is briefly described below with reference to Figure 2.

Such a density-responsive float element consists of a pivotally mounted buoyant element combined with a weight element in such a manner that the following criteria are fulfilled over the entire measuring range: 1. The radial line connecting the pivot with the centre of buoyancy has a higher angle (more positive) with reference to the horizontal than the radial line connecting the pivot with the centre of gravity, i.e., the float element is hydrostatically stable; 2. The density of the fluid (which the element is to measure) is greater than the density of the buoyant element, and less than the density of the weight element, i.e., if the fluid density is p, the density of the buoyant element Pa and density of the weight element Pb then the inequality of Pa < P < Pb is valid for all values of p.

The geometry of such a float element is shown in Figure 1, where the radial line connecting the pivot with the centre of buoyancy is displaced from the horizontal by an angle a and is separated from the radial line connecting the pivot with the centre of gravity by an angle ss. If the volume of the buoyant element is Va and its centre of buoyancy is at a radial distance A from the pivot, and the volume of the weight element is Vb and its centre of gravity is at a radial distance B from the pivot, the torque resulting from the couple of buoyant and gravitational force is: Equation (1) T = p(V,A cos a + VbB cos (a-)} - PaVaA cos a - PbVbB cos (ass) When immersed in a fluid, such an element will assume an angle at which torque is zero.The greater the relative density of the fluid, the higher (more positive) the angle a will be. This is a stable condition which is expressed by: (P - Pa) VaA cos a = (Pb - P) VbB cos (ass) (2) The angle which such an element will assume in a fluid of a given density may be taken from a rearrangement of (2), in which the first term describes the geometry of the float, the second term the relation of the densities, and the third term the corresponding angles:

The above expressions for length, volume, and density are assumed to be valid at some standard temperature. When subsequently considering the effect upon density measurements made at temperatures which differ from that standard temperature, it is obvious that all will be affected by thermal expansion.

When dealing with materials in which the coefficient of thermal expansion may vary with temperature, thermal expansion can be expressed by the dimensionless relation between a length L at the standard temperature and the change in this length AL at a different temperature 0. Thus, at temperature 0, any standard-temperature length L will be changed by the factor 1 + awl, and this changed length will be LH = L (1 +ALL}. Since AL is small, the resulting change in volume can be taken to be 3AL,so that the standard-temperature volume V will be changed to VH = V {1 + 3 3%L ).This volume change obviously results in an inverse density change, so that

If the thermal expansion ALL of the buoyant element is written simply Aa, and the thermal expansion of the .a element Ab and the element Ab, and the preceding expressions describing changes in length, volume, and density are introduced into equation (1), the angle which a float element will assume in a fluid of density Peat temperature0 may be taken from:

This equation is analogous to (3), and the fact that the first terms of each are identical is typical of float elements where, in agreement with previous practice, the weight element is structurally incorporated into the buoyant element.In this case, the thermal expansion of the buoyant material governs the change of both lengths A and B, so that these are proportional and therefore without effect on the thermal expansion characteristics of the float.

Since one of the purposes of the present invention is to provide improved temperature-error correction, structural means of influencing the thermal expansion characteristics of a float element will be proposed later. This requires the concept of an "attachment radius", which is the distance from the pivotal point of the float to the fixed attachment point of the buoyant element Ra and the corresponding distance for the weight Rb. The introduction of these terms avoids the restriction stated for (4), so that thermal expansion effects in any type of float are described by: Equation (5)

Equation (3) may be applied to any typical float element at standard temperature for deriving the changes in float angle corresponding to equal increments of fluid density; the result will be as shown in Figure 2.It will be seen that equal increments of density correspond to very unequal increments of float angle, and that this non-linearity limits the practically usable range of about 80 or less. This is an unavoidable characteristic of such devices, so that acceptable linearity in the measuring scale can be achieved only by restricting the total scale angle, i.e., by reducing either the measuring range, or the distance between scale subdivisions, or both.

As an example of a feature incorporated in the present invention, Figure 3 shows angular scales 1 and 2 corresponding to equal increments of fluid densities between 1020 and 1180 kg/m3 for a pair of float elements. Scale 1 applies to a float element 1 which has obviously been designed to respond with relatively large angular increments in the lower part of this density range, whereas scale 2 applies to a float element 2 which has been designed contrarily. Since the sector of greatest angular increments is the sector of greatest measuring accuracy, it may be said that float 1 responds predominantly in the lower part of the density range, and float 2 in the upper part.If each float is thus considered to predominate in half of the total range, it will be seen that each half-range encompasses an angle of 80 , shown by dashed lines, wherein the angular increments are not equal (it has been said that this is impossible to achieve), but are relatively widely spaced.

According to preferred embodiments of the present invention, one of these float elements is to carry a scale, and the other is to carry an index which can be read against the scale.

Figure 4 shows an embodiment corresponding to 1 of Figure 3, in the form of scale element 3, and an embodiment corresponding to 2 of Figure 3, in the form of index element 4. The two elements are free to rotate around a common axis. Scale element 3 carries a scale of angular increments corresponding to the relative angular displacement of the two elements at the specified densities, which can be confirmed as being the sums of the angular increments of the individual elements as shown in Figure 3. From Figure 4 it is apparent that the linearity of the resulting scale is comparatively excellent (being much better than that of the 80 half-ranges of Figure 3), and that, moreover, the scale has been expanded to encompass 210 (which is 30% more than the combined 80 half-ranges of Figure 3).The combination of float elements is therefore markedly superior to a simple "addition" of floats and associated scales, in regard to both scale linearity and total scale angle.

It will be noted that the scale of Figure 4 is symmetrical around the mid-point, and it may be correctly inferred that this is a consequence of the inverse symmetry of scales 1 and 2 of Figure 3. Obviously the scale angles shown in Figure 3 are in no way mandatory - by assigning a different angular range to the given density range, different angular increments would result - nor need the scales be symmetrical in any way. A salient advantage is to be seen in the fact that the angles assigned to each float element can be varied at will e.g., to expand or contract the resulting scale either as a whole or at any desired point, symmetrically or asymmetrically.

Figure 5 shows the embodiment of scale element 3 of Figure 4 in greater detail. if this element were a circular plate of uniform thickness, it would be hydrostatically neutral around the centrally located pivot 5. If it is made of buoyant material a hydrostatically effective buoyant volume (Va) can be achieved by increasing the thickness of the plate e.g., within the area 6. Angle a must then be taken from the radial line connecting the pivot with the centre of buoyancy of the added volume. Alternatively to increasing the plate thickness within area 6, the plate thickness can be reduced - or the plate material completely omitted - within the area 7, diametrically opposed to area 6, whereby the same effect is achieved.A hydrostatically effective buoyant volume can also be achieved by omitting some sector of the plate entirely, or by locating the pivot 5 off-centre of the circular plate of uniform thickness. In this example, a weight element 8 is structurally incorporated into the buoyant element, as is common practice.

Figure 6 shows the embodiment of index element 4 of Figure 4 in greater detail. The index element can of course be constructed according to the principles already described with reference to the scale element (e.g., by using a circular plate of small diameter, so that the scale is not obscured from view). However, any volume which is not symmetrical about the pivot point will be hydrostatically effective, so that it may be preferable to construct the index element as shown, whereby the buoyant element is a circular sector fitted with an index 9. It will be noted that the index 9 may be angularly displaced with reference to angle a, for instance to increase scale readability, but that of course the associated scale must be similarly displaced with reference to angle a of scale 3.In this example, the pivotal point 10 has the form of a ring which can rotate about the pivot 5 of scale element 3, so that both elements pivot a common axis. A weight element 11 is structurally incorporated into the buoyant element, as is common practice.

In Figures 4, 5 and 6 the two elements are so designed that the index element is positioned in front of the scale element. This order can naturally be reversed by positioning the scale element in front, as long as some index point remains visible, so that it can be read against the scale. It should also be noted that whereas scale element 3 is associated with the angular scale 1 of Figure 3, and index element 4 is associated with the angular scale 2 of Figure 3, this order is optional, and can be reversed without altering any of the principles involved or affecting any characteristics of the resulting density scale.

From the foregoing it is obvious that both float elements are free to assume angles to the horizontal in agreement with equation (3), whereby their relative angular position is an indication of the density of the fluid in which they are immersed. This is a further inherent advantage of this embodiment of the present invention, inasmuch as the density reading is completely independent of any type of "artificial horizon", as is mandatory in some previous devices of this type.

Since devices of this type are often used for density measurements at varying temperatures, the "temperature error" of such devices is important. The temperature error in angular degrees for any float element of typical construction is the difference in the angles a as calculated from equation (3) and equation (4). In previous practice, metallic weight elements have been used, and because of their small volume (high density) and low coefficients of thermal expansion, their contribution to temperature error is negligible. In such case, temperature error derives from the difference between the thermal expansion of the buoyant element and that of the fluid.Typical fluids (battery acids, anti-freeze solutions) have different coefficients of thermal expansion at different densities (the fact that these coefficients also vary with temperature will be discussed later), so that complete temperature compensation can be obtained for only a single fluid density.

This would be the case when the thermal expansion of the buoyant element is identical with that of the fluid.

Other densities of the same fluid, having other coefficients of thermal expansion, would still be subject to temperature error in all measurements not made at the standard temperature for which the device was designed. Under the above assumptions, temperature compensation is reduced to the selection of a buoyant material with a thermal expansion corresponding most closely to an "average" thermal expansion of the fluid within the measuring range of the instrument.

For a more ready understanding of thermal effects in device of this type, the thermal expansion typical of the materials in question can, for the sake of simplicity, be assumed to be a linearfunction of temperature change. Under this assumption, the thermal expansion curves of an idealized fluid would appear as straight lines, as shown in Figure 7 where the fluid density is assumed to vary from 1020 to 1180 kg/m3, and where the coefficient of thermal expansion of the fluid typically increases with increasing density. Curve 12 would then represent the thermal expansion of a bouyant material best "averaging" the thermal expansion of the fluid, and therefore providing the best possible correction of temperature error.

Figure 8 shows the resultant temperature error if the thermal expansion characteristics of Figure 7 are applied to a float element which is typically assumed to have a total scale angle of 80" within the assumed density range. The scale calibration is for the standard temperature shown in Figure 7, and the temperature range is from 0 C to +70 C. For selected densities, the temperature error at 00C is shown as a dashed line, and the corresponding temperature error at +70"C is shown as a dotted line. Temperature error has been fully corrected at fluid density 1100 kg/m3, inasmuch as here the thermal expansion of the buoyant material is identical with that of the fluid.Temperature errors are greatest at the extremes of the density range, and that the sign of this error is reversed from one extreme to the other is an obvious consequence of the relative thermal expansion shown in Figure 7.

Here it should be pointed out that angular differences resulting from thermal expansion are not actually a satisfactory indication of temperature error, since this would be more significant if expressed in terms of indicated density. Obviously, where scale subdivisions are widely spaced, a given angular difference represents a much smaller difference of indicated density than if the scale subdivisions at that point were narrowly spaced. Thus it may be concluded from Figure 8, that, although the angular errors at the extremes of the scale are not large, the errors of indicated density are in the order of + 5 kg/m3, and that this is to be taken as the maximum temperature error of the instrument.

The foregoing principles can be applied to the present invention with reference to the same idealized fluid, assuming the angles of the floats at standard temperature to be identical with those shown in Figure 3. If again, float 1 of Figure 3 is assumed to predominate in the lower part of the density range, i.e., from 1020 to 100 kg/m3, it would be analogous to the previous example to assume a buoyant material with a thermal expansion close to that of the fluid at density 1060 kg/m3. Conversely, float 2 should have nearly the same thermal expansion as that of the fluid at a density of 1140 kg/m3. This assumption is shown graphically in Figure 9, where curves 13 and 14 represent the thermal expansion of the buoyant materials of floats 1 and 2 respectively.

Figure 1 Oa shows the resultant temperature error for each float, whereas Figure lOb shows the resultant temperature error on the density scale, this total error being the sum of the angular differences resulting from thermal expansion effects. Here again, the error at 0 C is shown as a dashed line, and the error at +70"C as a dotted line, and these are seen to be greatest at the extremes of the scale. In this case, however, the error of indicated density is less, being nowhere greater than 13 kg/m3.

The summation of the temperature errors of the individual floats will be seen to follow a simple "rule of signs" which can be really understood in connection with FigureS: If the thermal expansion of the fluid lies between the thermal expansions of the individual floats, the angular error associated with the individual floats will have contrary signs, and the sum of the angular errors will be less than the greatest individual errors; if the thermal expansion of the fluid is greater (or less) than the greatest (or least) thermal expansions of the individual floats, the individual errors will have like signs, and their sum will be greater than the greatest individual error.In the first case, a combination of float elements according to the invention will result in an absolute reduction of angular error due to thermal expansion, whereas in the second case the opposite will be true.

In view of the above-mentioned "rule of signs", which is peculiar to the present invention, it is instructive to consider an example in which the total thermal-expansion range of the fluid lies between the thermal expansions of the individual floats, as shown in Figure 11, where curve 15 represents the thermal expansion of float 1, and curve 16 the thermal expansion of float 2. In this case, the thermal expansion of the individual floats does not correspond to that of the fluid at any density, and Figure 12a (which is analogous to Figure 1 Oa) shows that the angular differences corresponding to the temperature error of the individual floats are relatively large.In accordance with the "rule of signs", however, their summation as shown in Figure 1 2b is much smaller than the individual errors, so that the errors of indicated density are nowher greater than 12.5 kg/m3. Moreover, Figure 12b shows full temperature compensation, not only at the midpoint of the scale, but now also near both extremes of the scale, where the individual angular differences are very nearly equal in value, but contrary in sign.

The foregoing examples demonstrate the fact that in a device according to the present invention, buoyant materials having relatively wide-ranging coefficients of thermal expansion may be used, and yet the temperature compensation achieved by the device will be superior to what can be achieved with an ideal material in devices of previous type.

If, in the case of the present invention, ideal buoyant materials are assumed (analogous to the assumption of an ideal material for a conventional instrument as shown in Figures 7 and 8), the thermal expansion curves of floats 1 and 2 would appear approximately as shown by 17 and 18 in Figure 13. Figure 14a (which is analogous to Figures 10band 126) shows the total resultant temperature error, which ispractically negligible, being nowhere greater than 11 kg/m3 over the temperature range from 0 C to +70"C. Further, it will be seen that temperature error is completely compensated at fluid densities near 1040, 00, and 1160 kg/m3.Because the total angular differences are so small, the actual values are presented in Table I, where Aa is the angular error of the individual float, SZxa is the summation of the angular errors of both floats at the stated temperature, and Aa is the total angular differences over the entire temperature range. Table I illustrates how the contrary signs of the individual angular errors Aa result in an absolute reduction of temperature errorE:Aa at all densities and temperatures.

Standard Temperature of density measurement fluid 0 C +70"C density float 1 float 2 float 1 float 2 Aa Aa Sha Aa Aa Ea 1020 -0.7 +0.1 -0.6 +1.3 -0.2 +1.1 1.7 1060 -1.1 +1.7 +0.6 to1.9 -2.7 -0.8 1.4" 1100 -1.7 +1.6 -0.1 +3.0 -2.8 +0.2 0.3 1140 -1.8 +1.1 -0.7 +3.2 -2.0 +1.2 1.9 1180 0.0 +0.8 +0.8 +0.2 -1.5 -1.3 2.10 TABLE I If the sum of the angular errors at each temperature extreme (E::a) is related to angular increments of density at the corresponding points of the scale, the maximum temperature error of the instrument can be expressed in percent of indicated density, which is more relevant than angular degrees. Figure 14shows this error graphically for the entire density range, the temperature error at 0 C being shown as a dashed line and that at 70"C as a dotted line. It is obvious that temperature compensation is greatly enhanced by "S"-shaped error curves, which are perculiar to the present invention.

It was stated that the linearized thermal expansion curves used for the scale of simplicity in the foregoing comparative examples are idealized depictions of the properties of real materials. It should be pointed out that the validity of conclusions drawn from the examples is in no way dependent upon the linearity of thermal expansion curves as such, but rather upon the degree of divergence between curves pertaining to real fluids and curves pertaining to real float materials. In practice it will be found that the thermal expansion curves e.g., of a wide range of thermoplastics (which because of their density are well suited for use as a buoyant material) have characteristics similar to the thermal expansion curves of many aqueous solutions (.e.g., battery acid, glycol anti-freeze, etc).In both cases, the curves are nonlinear, showing an increase in coefficients of thermal expansion with increasing temperature, and this agreement permits the designer to select a buoyant material that will provide adequate temperature compensation for a given fluid in a given range of densities and temperatures. It was stated that temperature compensation in devices of this type is achieved primarily through selection of a buoyant material with suitable thermal expansion; thus in the present invention, the demands upon properties of materials are no more stringent than in devices of previous type.On the contrary, the foregoing examples of the invention, which assume three different pairs of buoyant materials (i.e., six different coefficients of thermal expansion), proved in every case to provide better temperature compensation than that achieved with an ideal material in a single-float instrument.

In order to exploit more fully the particular advantages of the present invention, it is a further purpose of the invention to provide new means of influencing the thermal expansion characteristics of the float elements. It was stated that in practice hitherto, metallic weight elements are used, and that their contribution to the thermal expansion characteristics of the float element as a whole is negligible. Contrary to this practice, it is within the scope of the present invention to utilize weight elements of relatively large volume and relatively large coefficients of thermal expansion, such that they contribute materially to the thermal expansion characteristics of the float element in such a way that these latter are no longer determined exclusively by the thermal expansion of the buoyant material. Obviously, such a weight element must satisfy the inequality Pb > P for all values of p.

The extent to which such a weight element as a whole may be judged by evaluating the first term of equation (4). If the radii A and B are assumed to be equal, the term is reduced to Vb/Va, and describes the relative volumes of the weight element and the buoyant element. In previous devices the metallic weight elements, this term would assume values - depending upon the given densities and associated float angles ranging roughly from 0.015 to 0.04, i.e., the volume of the buoyant element would be 25 to 70 times greater than that of the weight element.

However, if the relative volumes are more nearly equal, the designer can utilize the thermal expansion of the weight element to modify the thermal expansion characteristics of the float element as a whole. This is accomplished in the present invention by using a weight element of relatively low density, such that the density of the weight element is not greater than four times the density of the buoyant element, i.e. pb < 4. If, pa for example, the weight element is assumed to have a density about one and a half times greater than that of the buoyant element, the ratio VbIVa would assume values ranging roughly from 0.3 to 3.0.Therefore, a weight element of specified density might have one third of the volume of the buoyant element, or it might have three times the volume of the buoyant element, depending upon the density range and associated angles assigned to the float element. Thus, the designer, having chosen two suitable materials, may further influence the relative volumes of the two materials, and thereby the thermal expansion characteristics of the float element as a whole, by varying the density range and/or the angular range of the float in question. From this it will be seen that the advantage resulting from the use of a low-density weight element according to the invention is not merely "additive", but rather provides a new dimension of design freedom, in which temperature compensation is no longer determined by the choice of materials alone.

As a further new means of influencing the thermal expansion characteristics of any float element, the present invention proposes a float assembly in which a difference between the coefficients of thermal expansion of the buoyant material and the weight material will result in a disproportional change in the hydrostatic radii of the respective elements with change in temperature. It was stated that in previous float elements the weight element is structurally incorporated into the buoyant element, so that the thermal expansion of the latter determines changes in the length of both radii, A and B, and that such changes are therefore proportional and without effect upon the thermal expansion characteristics of the float element.

However, the utilization of relatively low-density (i.e., large-volume) weight elements as previously proposed in the invention makes it feasible to design float assemblies in which the relative angular positions of the buoyant element and weight element are maintained, while permitting different rates of change in the respective radii A and B, corresponding to the different expansion of the element with change of temperature. In this case, the float angles will be governed by equation (5), rather than by equation (4) as heretofore.

Figures 1 5a, band c show embodiments of float elements utilizing the foregoing means to modify thermal expansion characteristics.

Figure 1 5a shows an embodiment similar to index element 4 of Figure 4, where the buoyant element 19 is structurally incorporated into a weight element 20 of relatively low density. This is a reversal of previous practice, and, since the thermal expansion of the weight element could be expected to be relatively small, the thermal expansion of the float element would be markedly less than that of the buoyant material alone. Here, the thermal expansion of the weight element governs the change in both radii, A and B, so that the equation (4) still applies.

Figure 1 5b shows a similar embodiment, where a relatively low-density weight element 21 and a buoyant element 22 have a fixed attachment at the pivotal point by means of concentric rings 23 and 24. A sliding "dovetail" attachment 25 prevents any angular displacement of the elements relative to each other, while permitting different radial expansions with change of temperature. If the two materials have different coefficients of thermal expansion, changes of temperature will cause disproportional changes of radii A and B, so that equation (5) applies.Since the fixed attachment of both elements is at the pivotal point, the attachment radius for both elements, R0 and Rb, is zero, and the first term of equation (5) becomes:

Figure 1 Sc shows a similar embodiment, where the fixed attachment 28 of weight element 26 to buoyant element 27 is located at the periphery of the float, while a sliding attachment 29 prevents any relative angular displacement of the two elements. Here the buoyant element is likewise attached at the pivotal point, so that the radius of attachment R0 is zero. The weight element, however, is attached at 28, so that its radius of attachment may be said to have a value B + X.Substitution in the first term of equation (5) gives:

Therefore, although the volumes and materials of the floats described in 1 Sb and 1 5c might be identical, their thermal expansion characteristics would be different because of the structural modification of the respective attachment radii according to the invention.

Obviously, the foregoing principles may be applied to either or both of the float elements according to the present invention, and it has been said that their effects upon temperature compensation will depend upon the density range and corresponding angular range assigned to each float. Since a variety of suitable materials is available, and since the latter, as is well known, may be modified by the admixture of inert "fillers", it may be concluded that the particular advantages of the present invention can be fully exploited in practice.

By way of verifying the previously stated principle in applications with real materials, the case of a precision battery-acid tester may be considered as an example. The usual density range of such an instrument would be from 1100 to 1300 kg/m3, and the acid temperature could be assumed to range from +10'Cto +45"C. In this example, the buoyant material of float 1 is assumed to be an unmodified high-density polyethylene, the buoyant material of float 2 an unmodified low-density polyethylene, and the weight material of both floats an unmodified polyfluorocarbon. The attachment radii of the float and weight elements are both zero, and the total scale angle is 210 , as in the previous examples.

Figure 16 shows temperature error graphically in percent of indicated density at temperatures of +10"C (dashed line), +25"C (solid line), and +45"C (dotted line). Figure 16 fully confirms the conclusions drawn from the foregoing simplified assumptions, and demonstrates that by application of the principles of the present invention to real materials, temperature error can be reduced to an extent impossible with devices of prior type.

In this connection, it is important to consider the fact that, although the present device can be used simply by immersing it into a fluid the density of which is to be measured, more often such a device is made portable by enclosing it within a housing which is at least partly transparent, whereby some means (usually a compressible bulb) is provided for drawing a fluid sample into the housing in such a way that the float elements are immersed. In practice, it has been found that density measurements made with such portable devices are subject to serious errors, for the simple reason that the seemingly unavoidable air bubbles in the fluid sample tend to attach themselves to the float element, where their added buoyancy caused significant error in the float angle.It is necessaryforthe fluid inlet of such a housing to be small, since it is surface tension within this orifice which prevents involuntary outflow of the fluid sample once the inlet orifice is no longer immersed in the body of fluid from which the sample was drawn. Furthermore, a compressible bulb with sufficient elasticity to fill the housing within a reasonable time will cause a high pressure gradient at the inlet during the beginning of fluid inflow, when the bulb is most compressed. Also, at this time the weight of the fluid supported by the pressure gradient will at a mininum. As a result, the inflow velocity will be high; flow speeds of 3 m/sec and more are not uncommon.In order to ensure that this jet of fluid will not impinge upon the elements contained within the housing, it is customary to provide the latter with interior deflectors of various forms. However, at such speeds the fluid jet disintegrates, regardless of where it impinges, and as a result the fluid samples will contain air bubbles, which may falsify the density measurement. For this reason, the user of such instruments is always instructed to tap them in order to remove air bubbles before taking a density reading.

This has been unavoidable, because a significant reduction of inflow speed - achieved, for example, by a reduction of bulb elasticity - would result in an intolerable prolongation if inflow time. The user of such an instrument would then tend to withdraw the inlet opening from the body offluid from which the sample was being drawn before the bulb was fully expanded. In consequence, further expansion of the bulb would draw air bubbles into the housing at the end of fluid inflow, although such bubbles might have been avoided at the beginning.

Since this falsification of density measurements is a serious deficiency in devices of this type, it is a further purpose of the present invention to provide a means for preventing the formation of air bubbles in the fluid sample under normal conditions of use. To this end, the float elements as previously described are enclosed in a known housing which is at least partly transparent and which is provided with a known means for drawing a fluid sample into the housing in such a manner that the float elements are immersed in the sample.The fluid inlet consists of a small exterior orifice which, according to the invention, is connected to the chamber containing the float elements by a channel of gradually increasing cross-section area, such that the area of cross-section at the point of entrance into the chamber is materially larger than the cross-section area of the exterior orifice. Since the quantity of fluid passing the exterior orifice at any given time must be equal to the quantity entering the housing chamber, the flow speed at these points will be inversely proportional to the area cross-section.

The rate of deceleration within this channel will be limited by the fact that the resulting pressure gradient must be moderate if the fluid stream is not to become dissociated by the formation of air bubles, which would be detrimental to the effect desired. Inflow speeds at the exterior orifice will of course vary with orifice size, and with the pressure gradient producing the inflow, but typical dimensions would result in Reynolds numbers not greater than 1 104. Under such conditions, it may be said that - depending upon the size of the exterior orifice and the arrangement of elements within the housing chamber - the largest cross-section area ofthe channel should be at leastfourtimes greater than that of the exterior orifice, and that a conical inlet should have a half-angle of not more than 5 .

However, a conical inlet channel of circular cross-section is not mandatory, and under certain circumstances it might be advantageous to utilize a divergent channel of rectangular or other cross-section form. In such cases, reference can be made to a critertion of similarity based on the circumference C and the area A of the channel cross-section. (This is analogous to the "hydraulic radius" A/C, which is a criterion of similarity for fluid flow in straight pipes). If dA is an increment of area corresponding to dL, an increment of length in the direction of flow, divergent channel may be considered similar if the expression 1/C dA/dL is similar. For a conical channel - in view of the small angles of divergence in question - this expression reduces to dr/dL, where r is the cross-section radius.The preceding specification can therefore be restated for a conical channel in the form: < dL < 0.09 and for any divergent channel according to the invention in the form: 0 < 0. dl < 0.09 If the advantage resulting from utilization of a divergent inlet channel is to be retained under all practical conditions of use, it is necessary to consider the fact that in devices of this sort, the inlet is often provided with a tubular extension of some length suitable to a particular application. In such cases it is common practice to attach a length of flexible tubing which fits over the inlet orifice of the device.Since any significant length of tubing with very small diameter would increase inflow resistance and prolong inflow time, such tubing is almost always of materially larger inside diameter than the inlet orifice of the device itself.

Because of this, and because of uncontrolled movements of the flexible tubing, fluid contained in the tubing is often lost by involuntaty outflow once the open end of the tubing is no longer immersed in the fluid from which the sample is being drawn. Although the amount of fluid thus lost is small, the height of the fluid column represented by the tubing is large compared with the height of the device as a whole. In consequence, the pressure equilibrium existing within the system at the end of fluid inflow is significantly disturbed by any loss of fluid from the tubing; the compressible bulb expands futher and air is drawn into the device, even though there may be no loss of fluid through the inlet orifice itself.In the case of a very flexible compressible bulb, the slightest disturbance of pressure equilibrium can initiate bulb expansion, which in turn shortens the fluid column in the tubing, thus accelerating the process of aspiration whereby first fluid, then air, is drawn from the tubing into the housing chamber. This process may even assume an oscillatory character, whereby air intake alternates repeatedly with fluid loss.

Because of the importance of avoiding air bubbles and turbulence within the chamber containing the float elements, the invention provides further that the inlet channel be arranged in the form of an inverted "U"-tube, the highest point of which is located substantially higher than the point where the channel enters the housing chamber. By this means, air which passes the exterior orifice of the channel after the chamber is filled is entrapped at the highest point of the channel, whereby a hydrostatically stable condition is established, so that there is no airbleed into, or loss of fluid from, the housing chamber itself.

Because the space available for accommodating the inlet channel as described would be limited in practice, it is preferable to provide a relatively greater degree of divergence in that part of the channel which is between the exterior orifice and the highest point, and a relatively lesser degree of divergence in that part of the channel which is between the highest point and the entrance of the channel into the housing chamber.

By this feature, the negative acceleration of the fluid column which is induced by the divergence of the channel is greater where it is augmented by a negative gravitational acceleration, and relatively less where a positive gravitational acceleration, being contrary to that induced by the divergence of the channel, increases the risk of dissociation of the fluid column. By this means a maximum reduction of inflow speed may be achieved for any given channel length.

Figure 17 shows an example of a portable density-measuring device according to the invention. The transparent housing 30 contains the coaxially mounted pivoting float elements 31 and 32, and is fitted with a compressible bulb 33. The compressible volume of bulb 33 is such that the housing 30 can be completely filled with a fluid sample so that the float elements 31 and 32 are immersed in the latter. The fluid sample is admitted into the housing at the exterior orifice 34, from where it passes through channel 35 of gradually increasing cross-section area and enters the housing chamber at 36 with a materially reduced inflow speed.

Channel 35 follows roughly the shape of an inverted "U"-tube, the highest point of which is located higher than point 36, where it enters the housing chamber. A suitable length of tubing 37 can be attached at exterior orifice 34.

Fluid density can also be measured by means of hydrometers which are traditionally made of glass. Since the coefficient of thermal expansion of glass is much smaller than that of most fluids, density measurements made at different temperatures with conventional glass hydrometers are subject to large temperature errors.

This has led to the proposal of hydrometers made, for instance, of molded plastics which have relatively larger coefficients of thermal expansion. Such plastics hydrometers have been proposed either with a ballast means to provide additional mass and to encourage the device to float upright, or without such ballast means, in which case the hydrometer is guided in an upright position inside a vertical tube or channel wherein the hydrometer is buoyant in the fluid. In conjunction with this, it has been proposed to make the hydrometer of a material which has substantially the same coefficient of thermal expansion as that of the fluid, so that temperature error will be minimized.

As has already been stated the thermal-expansion characteristics of most fluids reveals that these have different coefficients of thermal expansion at different densities and thus this last proposal only achieves full compensation at a single temperature.

For a more ready understanding of thermal effects in devices of the present type, the thermal expansion typical of the material in question can, for the sake of simplicity, be assumed to be a linearfunction of temperature change. Under this assumption, the thermal expansion curves of a typical fluid would appear as straight lines as shown in Figure 18, where the fluid density is assumed to vary from a minimum density Pmin with a coefficient of thermal expansion At (Pmin) to a maximum density Pmax with a coefficient of thermal expansion a (Pmax). The thermal expansion curves of the intermediate fluid densities lie between these extremes.Curve 51 would then represent the thermal expansion of the hydrometer material which best "averages" the thermal expansion of the fluid within the given density range, thereby providing the best possible temperature compensation. From Figure 18 it is apparent that complete temperature compensation is achieved only in the middle of the density range, and that at extreme temperatures and densities, indicated by 52, 53, 54 and 55, temperature error is unavoidable. This temperature error obviously results from discrepancies between the thermal expansion of the fluid and that of the hydrometer at any given temperature and density, and its magnitude is such that the reliability of the device can be significantly impaired.

The present invention provides means for correcting the temperature error, as can be understood with reference to Figure 19. In Figure 19, the total volume of the hydrometer is divided into three significant volumes, here designated as the body volume Vx, which is the volume of the body of the hydrometer up to the lowest marking of the hydrometer scale; the scale volume Vy, which is the volume between the lowest and the highest marking of the hydrometer scale; and the stem volume Vz, which is the remaining volume above the highest marking of the hydrometer scale. When the hydrometer is buoyant in a fluid at the highest part of the density range, only Vx is immersed in the fluid. When the hydrometer is buoyant at the lowest part of the density range, both volumes Vx and Vy are immersed in the fluid.Volume Vz is necessary in order to permit accurate readings at the lowest fluid density, and, as will be explained later, the size of this volume can serve to regulate the mass of the hydrometer, so that volumes Vx and Vy can be accurately predetermined.

According to the invention, volume Vx is comprised of a material or combination of materials such that the coefficient of thermal expansion of volume Vx is substantially the same as that of the fluid at the highest part of the density range. At the time, volume By is comprised of a material or a combination of materials such that the coefficients of thermal expansion of the combined volumes Vx and Vy is substantially the same as that of the fluid at the lowest part of the density range.In terms of the previously stated expressions for the density and thermal expansion of the fluid, and writing the thermal expansion of volume Vx simply as AX and that ofvolume Vy simply as At the above conditions are fulfilled when

Figure 20 shows the resulting thermal expansion curve 6 for volume Vx and curve 7 for the combination of volumes Vx and Vy in relation to the thermal expansion curves of the fluid as shown in Figure 18. Because of the coincidence of the curves at both the highest and lowest densities, temperature error is avoided.At densities intermediate between these extremes, the thermal expansion of the hydrometer is governed by the relation between volume Vx and the immersed portion of volume Vyt and is therefore proportional to the density of the fluid. Since the same proportionality applies substantially to the thermal expansion of the fluid at intermediate densities, the thermal expansion of the hydrometer is at all times analogous to that of the fluid, so that temperature error is avoided through the entire measuring range of the instrument.

Figures 21a and 21b show an example of an embodiment of the invention in the form of a moulded plastics hydrometer which is composed of two different materials. Volume Vy consists of a material 59, whereas volume Vx is formed by a structural combination of material 59 with a material 58. Since a circular cross-section is in no way mandatory to the functioning of hydrometers, the cross-section of volume Vx might be arranged as shown in Figure 21 b, in order that both materials may assume the temperature of the fluid in which the hydrometer is immersed within a short time.

In order to achieve full temperature compensation in a fluid as previously characterized, the thermal expansion of material 58 should be greater than that of the fluid at highest density, whereas the thermal expansion of material 59 should be less than that of the fluid at lowest density. This assumption is shown graphically in Figure 22, where curve 58 indicates the thermal expansion of material 58, and curve 59 that of material 59. At the highest fluid density, volume Vx is immersed, and it may be composed of 65% material 58 and 35% material 59. The thermal expansion of volume Vx would therefore lie proportionally between those of materials 58 and 59, as shown by curve 60. At the lowest density, both volumes Vx and Vy would immersed, and since Vx consists solely of material 59, the relative proportion would be 40% material 58 and 60% material 59.Here again, the immersed volume would have a thermal expansion reflecting this proportion, as shown by curve 61. Obviously, similar results would be obtained by using other materials with other coefficients of thermal expansion in different proportions.

It was stated that the linearized thermal-expansion curves used to illustrate the foregoing examples are idealized depictions of the properties of real materials. The conclusions drawn from the examples are, however, quite independent of the linearity of thermal expansion. In practice, it will be found that the thermal-expansion curves of a wide range of suitable hydrometer materials are similar in character to the thermal-expansion curves of most aqueous solutions. In both cases, the curves show an increase in the coefficients of thermal expansion with increasing temperature, and because of this similarity, numerous suitable materials can be combined in accordance with the foregoing principles for use with different fluids.

In practical applications of the present invention, it is obvious that the effective size of volume Vx and Vy will depend upon the actual depth of immersion ofthe hydrometers Since immersion is governed by the mass of the hydrometer, this must be accurately predetermined. Offhand, it might be said that hydrometer mass M could be taken from a simple buoyancy calculation, for example M = Vxpmax = (Vx + Vy) Pmin but in practice it will be found that surface tension acting upon buoyant hydrometer will cause an increase of immersion depth which would significantly alter the required volumetric relations. For this reason, it is within the scope of the present invention to provide means for accurately compensating for surface-tension effects, so that the hydrometer will float at a predetermined depth of immersion.

If the surface tension of the fluid is o(N/m), the resulting increase in depth of immersion Ad can be taken as a good approximation from d = 9 81 0,where N is the circumference and Q the area of the hydrometer cross-section at the level of the fluid surface. Since Ad is a function of fluid density, it is at present more convenient to assume the increased immersion to result from an apparent additional mass m = gN8" which is independent of the variable density term. This may be thought of as being the mass of fluid which is raised above the free surface by capillary attraction to the hydrometer.For the purpose of buoyancy calculations, the total mass of the hydrometer would than be defined as M = M > c + My + M2 + gN8" where Mx, My, and M2 are the mass of volume Vx, Vy, and V2, respectively. In order that the advantages of the present invention may be fully realized in practice, the hydrometer also comprises a stem volume V2 of density pz, which is so dimensioned that aN )1 (8) V, = (V,p,,, ) Pz It has been said that hydrometers may be either ballasted or unballasted. In the case of the former, the density of the materials is not often critical, since the total mass of the hydrometer can be controlled by varying the mass of the ballast.In the case of the latter, the total mass would be controlled by dimensioning volume V2 in agreement with equation (8). The extent of this means is, however, limited by the fact that the unballasted hydrometer would be excessively "top-heavy", and therefore subject to increased friction within the guiding tube in which it is buoyant, if volume V2 were to be excessively enlarged. (A certain minimum volume Vc is of course necessary in order to prevent inadvertent total immersion of the hydrometer in the lower part of the density range).In this case, the density of hydrometer materials would be critical, and it remains to be demonstrated that the foregoing principles can be put into practice under stringent conditions where real materials must simultaneously satisfy both density and thermal-expansion requirements in applications with real fluids.

The most common example would be density measurements involving sulphuric-acid solutions, where the corrosiveness of the fluid practically precludes the use of metallic ballast. Measurements made with glass hydrometers are subject to large temperature errors, but unballasted hydrometers of molded polystyrene can be used, since this material has an appropriate density, a larger coefficient of thermal expansion, and sufficient chemical resistance.The most frequently used density range is from 1100 kg/m3 to 1300 kg/m3, and Figure 23 shows to scale an appropriate polystyrene hydrometer of density 1050 kg/m3. It will be noted that the stem volume V2 is barely sufficient to ensure functioning of the hydrometer at the lowest fluid density, but less dense materials (e.g., polyethylenes) would require a much larger volume V2, as shown by contour 62, and would therefore be unacceptable. Thus, Figure 23 represents an optimal hydrometer for this case according to the prior art.

Figure 24 shows to the same scale an embodiment of the present invention for this case, and having volumes Vx and Vy identical with those of the hydrometer in Figure 23. Volumes Vy and V2 consist of SAN (styrene acrylonitrile) copolymer of density 1080 kg/m2, whereas volume Vx is a combination of 53% SAN and 47% polyethylene of density 917 kg/m3. Here it will be seen that the resulting volume Vz is well-proportioned, and that therefore this combination of materials satisfies the density requirements fully.

Figure 25 shows the temperature error of each of these hydrometers as a percent of indicated density when the fluid temperature is assumed to vary from 10 to 60"C. Although the temperature error of the hydrometer of Figure 23 is less than that of a glass hydrometer, it will be seen to vary from a maximum of about -0.8% at 10 C (curve 13) to +0.8% at 60"C (curve 64). The corresponding curves 65 and 66, which apply to the hydrometer of Figure 24, show a temperature error nowhere greater than +0.2%, i.e. an improvement in temperature compensation by factor of 4.

In the foregoing example, Figure 24 represents the simplest possible embodiment of the invention, i.e., an unballasted hydrometer consisting of only two materials. In cases where temperature compensation would be subject to very stringent requirements, it would of course be within the scope of the present invention to combine more than two materials in such a way that the thermal expansion of the hydrometer would be in even closer agreement with heat of the fluid. As an example, Figure 26 shows thermal-expansion curves of a fluid where the change in coefficients of thermal expansion is disproportional to the change in density, i.e.

where thermal expansion decreases more rapidly with decreasing density. For this case, a preferred embodiment of the invention could be that shown in Figure 27, where volume Vy comprises a combination of a material 67 with a material 68. In order for the thermal expansion of the hydrometer to be analogous with that of the fluid as characterized in Figure 26, the thermal expansion of material 68 would be less than that of material 67, whereby the thermal expansion of the immersed volume of the hydrometer would also decrease more rapidly with decreasing density.

It will be understood that the constant 9.81 used herein designates the acceleration of gravity (g) in metres - kilogram - second system, and is dimensionally compatible with surface tension in Newtons/metre. When using other systems, appropriate different gravity acceleration constants should be used.

It will be appreciated that the hydrometer could be disposed in a housing of the general type hereinbefore described and/or used with a fluid inlet construction described. The invention includes such a housing and or the fluid inlet in construction per se.

Claims

1. A device for measuring the density of a fluid having a density within a predetermined range and a coefficient of thermal expansion that varies over this range, comprising a body formed such that when floated in the fluid it gives an indication of the fluid density and having an effective coefficient of expansion which varies over the predetermined range in a sense to compensate for variations in the coefficient of thermal expansion ofthefluid.

2. A device as claimed in claim 1 wherein the body is formed in first and second parts.

3. A device for measuring the density of a fluid having a density within a predetermined range, comprising a body formed such that when floated in the fluid it gives an indication of the fluid density, the body comprising two parts, each having respective buoyancy characteristics, the arrangement being such that the buoyancy characteristic of the first of the parts predominates at the lower end of the predetermined range and the buoyancy characteristic of the second part predominates at the upper end of the predetermined range.

4. A device as claimed in claim 2 or claim 3 wherein the first and second parts are relatively rotatable about an axis.

5. A device as claimed in claim 4 wherein each part has a centre of bouyancy and a centre of gravity offset from the axis and from each other.

6. A device as claimed in claim 4 or 5 wherein one of the parts carries a scale and the other a pointer.

7. A device as claimed in anyone of claims 2 to 6 wherein at least one of the first and second parts is made in two sections having different coefficients of thermal expansion.

8. A device as claimed in claim 7 wherein the two sections are dovetailed or are pivotally interconnected.

9. A device as claimed in anyone of claims 2 to 8 where in at least one of the parts includes a weight, a cutout or an indentation.

10. A device as claimed in claim 2 or claim 3 wherein the first and second parts are integral or fixed one to the other, and wherein the second part extends above the first when the body floats.

11. A device as claimed in claim 10 wherein the mean coefficients of thermal expansion of the two parts are different.

12. A device as claimed in claim 8 or claim 9 wherein the mean coefficient of thermal expansion of the second part varies along its length in the direction of immersion.

13. A device as claimed in anyone of claims 10 to 12 wherein the second part carries a scale.

14. A device for measuring the density of a fluid having a density within a predetermined range, comprising a housing for containing the fluid, a first float element pivotally attached to a horizontal axis in said housing, a second float element pivotally attached to said axis, each of said elements including at least one buoyant volume that is less dense relatively to the elements considered as a whole and has a centre of buoyancy displaced from said axis, said buoyant volume having a density which is less than the fluid density to be measured and at least one weight volume that is more dense relative to the element considered as a whole and has its centre of gravity displaced from said axis and a density greater than the fluid density to be measured, each of said elements being formed so that the radius joining said centre of buoyancy to said axis is angularly displaced from and above the radius joining said centre of gravity to said axis, whereby because of differing buoyancy in fluids of different densities the relative angular positions of said elements about said axis indicate the density of the fluid in which they are immersed and said elements rotate in opposite directions about said axis with changes in the density of the fluids, said first float element being constructed and arranged to respond to equal increments of fluid density within said predetermined range with relatively larger angular increments of rotation about said axis in a lower section of said density range, said second float element being constructed and arranged to respond to equal increments of fluid density within said predetermined range with relatively larger angular increments of rotation about said axis in a higher section of said density range, so that the relative angular displacement of said elements in said lower section of said density range, is determined to a greater extent by said first element than by said second element and, in said higher section of said density range, to a greater extent by said second element than by said first element.

15. A device as claimed in claim 14, arranged to measure the density of a fluid having a density that within said range varies with the temperature of the fluid, and having thermal expansion characteristics that vary with density within said range, said first float element being formed so as to relatively approximate the thermal expansion of the fluid when the fluid is in a lower section of said range, and said second float element being formed so as to relatively approximate the thermal expansion of the fluid when the fluid is in a higher section of said range, whereby deviations in the angular positions about said axis of said first element due to temperature variations of the fluid are relatively small in said lower section of said density range, and deviations in the angular positions about said axis of said second element due to said temperature variations are relatively small in said higher section of said density range.

16. A device as claimed in claim 15 wherein said float elements are formed from materials such that the difference between the average thermal expansion of the fluid and the thermal expansion of said first element is of approximately the same magnitude as, but opposite in sign to, the difference between the average thermal expansion of the fluid and the thermal expansion of said second element, whereby at any point in said density range deviations in the angular positions of said first element about said axis resulting from thermal expansion effects are of approximately the same magnitude as, but opposite in direction to, the corresponding deviations in the angular positions of said second element, the relative angular displacement of said elements about said axis resulting from thermal expansion effects thereby being of smaller magnitude than said deviations.

17. A device as claimed in anyone of claims 14to 16, arranged to measure the density of fluids having thermal expansion characteristics that vary within density within said range, said float element being so formed that the thermal expansion of the fluids over said range lies between the thermal expansions of said float elements.

18. A device for measuring the density of fluid having a density within a predetermined range, comprising a housing for containing the fluid, a first float element pivotally attached to a horizontal axis in said housing, a second float element pivotally attached to said axis, each of said elements including at least one buoyancy volume that is less dense relative to the element considered as a whole and has a centre of buoyancy displaced from said axis, said buoyant volume having density which is less than the fluid density to be measured and at least one weight volume that is more dense relatively to the element considered as a whole and has its centre of gravity displaced from said axis and a density greater than the fluid density to be measured, each of said elements being formed so that the radius joining said centre of buoyancy to said axis is angularly displaced from and above the radius joining said centre of gravity to said axis, whereby because of differing buoyancy in fluids of different densities the relative angular positions of said elements about said axis indicate the density of the fluid in which they are immersed and said elements rotate in opposite directions about said axis with changes in the density of the fluids, at least one of said float elements consisting of a relatively more dense weight volume and a relatively less dense buoyant volume, the density of said weight volume being less than four times the density of said buoyant volume whereby the size of said weight volume relative to said buoyant volume whereby the size of said weight volume relative to said buoyant volume is such that the thermal expansion of said weight volume contributes materially to the thermal expansion of said float element.

19. A device for measuring the density of a fluid having a density within a predetermined range, comprising a housing for containing the fluid, a first float element pivotally attached to a horizontal axis in said housing, a second float element pivotally attached to said axis, each of said elements including at least one buoyant volume that is less dense relatively to the element considered as a whole and has a centre of buoyancy displaced from said axis, said buoyant volume having a density which is less than the fluid density to be measured and at least one weight volume that is more dense relatively to the element considered as a whole and has its centre of gravity displaced from said axis and a density greater than the fluid density to be measured, each of said elements being formed so that the radius joining said centre of buoyancy to said axis is angularly displaced from and above the radius joining said centre of gravity to said axis, whereby because of differing buoyancy in fluids of different densities the relative angular positions of said elements about said axis indicate the density of the fluid in which they are immersed and said elements rotate in opposite directions about said axis with changes in the density of the fluids, at least one of said float elements consisting of a buoyant volume having a centre of buoyancy and a weight volume having a centre of gravity which is angularly displaced from said centre of buoyancy said volumes having a common pivotal point concentric to said axis and being constrained against relative angular displacement of said centres but unconstrained against individual radial thermal expansion with respect to said axis, such that said angular displacement of said centres is maintained when said volumes undergo individual radial thermal expansion with respect to said axis resulting in a mutually independent radial displacement of said centre of buoyancy relative to said centre of gravity which is governed solely by the individual coefficients of thermal expansions of said volumes.

20. A device for measuring the density of a fluid having a density within a predetermined range, comprising a housing for containing the fluid, a first float element pivotally attached to a horizontal axis in said housing, a second float element pivotally attached to said axis, each of said elements including at least one buoyant volume that is less dense relatively to the element considered as a whole and has a centre of buoyancy displaced from said axis, said buoyant volume having a density which is less than the fluid density to be measured and at least one weight volume that is more dense relatively to the element considered as a whole and has its centre of gravity displaced from said axis and a density greater than the fluid density to be measured each of said elements being formed so that the radius joining said centre of buoyancy to said axis is angularly displaced from and above the radius joining said centre of gravity to said axis, whereby because of differing buoyancy in fluids of differing densities the relative angular positions of said elements about said axis indicate the density of the fluid in which they are immersed and said elements rotate in opposite directions about said axis with changes in the density of the fluids, at least one of said float elements consisting of a buoyant volume having a centre of buoyancy and a weight volume having a centre of gravity which is angularly displaced from said centre of buoyancy, one of said volumes being attached pivotally to said axis and the other of said volumes being attached to said one of said volumes at a point radially distant from said axis for individual thermal expansion relative to said one of said volumes, said other volume being thereby attached indirectly to said axis and arranged for said angular displacement of said centres to be maintained, whereby changes of temperature affecting said volumes result in radial displacement of said centre of said indirectly attached volume governed by the coefficient of thermal expansion of said pivotally attached volume and by the coefficient of thermal expansion of said indirectly attached volume affecting the radial displacement of said centre of said indirectly attached volume relative to said pivotally attached volume.

21. A device as claimed in anyone of claims 14 to 20 wherein said housing is at least partially transparent and the device also includes suction means positioned for drawing a fluid sample into said housing so as completely to immerse said float elements, and duct means formed with a divergent channel for admitting said fluid sample into said housing, said channel having an inlet orifice located outside said housing at which the cross-sectional area of said divergent channel is smallest, and said channel so gradually increasing in cross-sectional area that the increment of cross-sectional area per increment of length at any point of said channel forms a ratio with the circumference at that point which is not greater than 0.09, the divergent length of said channel being such that the cross-sectional area increases at least by a factor of 4.

22. A device as claimed in claim 21 wherein said divergent channel is formed as an inverted U-tube located with its highest point substantially higher than the point at which said channel enters said housing.

23. A device as claimed in claim 22 wherein said channel has a relatively greater degree of divergence in that part of the channel which is between the exterior inlet orifice and the highest point, and a relatively lesser degree of divergence in that part of the channel which is between the highest point and the entrance of the channel into the housing chamber.

24. A device anyone of claims 14 to 23 wherein one of said float elements is marked with a scale corresponding to fluid density and the other of said float elements includes an index or pointer adjacent to said scale, said scale and said index or pointer being located to provide an indication of fluid density on said scale in all effective relative positions of said float elements within said predetermined density range.

25. A hydrometer for measuring the density of a fluid having a density within a predetermined range and a coefficient of thermal expansion that varies within this range, comprising a lower body volume Vx which is equal to the total mass of the hydrometer divided by the density of said fluid at the highest part of said range, and an intermediate scale volume Vywhich together with said volume Vx is equal to the total mass of the hydrometer divided by the density of said fluid at the lowest part of said range, and an upper stem volume Vz whereby said volume Vx is comprised of one or more materials so proportioned that the coefficient of thermal expansion of said volume Vx is approximately equal to that of the fluid at the highest part of said range, and said volume Vy is comprised of one or more materials so proportioned that the coefficient of thermal expansion of the combined said volumes Vx and Vy is approximately equal to that of said fluid at the lowest part of said range.

26. A hydrometer as claimed in claim 25, wherein the mass of said stem volume Vz, when added to the combined masses of said volumes Vx and Vy, is equal to the product of said volume Vx and the density of said fluid at the highest part of said range, less the apparent additional mass resulting from the surface tension of said fluid, said apparent additional mass being calculated as the product of the surface tension of the fluid and the circumference of the hydrometer at the level of the fluid surface, divided by the acceleration of gravity.

27. A hydrometer as claimed in claim 25 or claim 26 for use with fluids whose coefficients of thermal expansion change disproportionally with density, wherein said scale volume Vy is comprised of at least two materials which have different coefficients of thermal expansion and which are unequally distributed within said volume Vy, so that the lower part of said volume Vy has a coefficient of thermal expansion different from that of the higher part of said volume Vy.

28. A device for measuring the density of a fluid having a density within a predetermined range, substantially in any of the forms hereinbefore described with reference to the accompanying drawings.

29. A device for measuring the density of a fluid having a density within a predetermined range, comprising a body formed such that when floated in the fluid it gives an indication of the fluid density, different portions of the body having different physical characteristics such that the characteristics of one portion predominates in determinating the indication at one end of the range and that the characteristics of the other portion or a combination of the portion predominates in determining the indication at the other end of the range.