the same factory. One creamery using farm separator cream exclusively, used the oil-test churn entirely at the time inquiry was made. The rather surprising fact that many are not buying cream by the Babcock test, even where they are buying milk by it, shows there are difficulties in its use when applied to cream.

It is evident that when the same price per pound is paid for butter showed by the oil-test churn, as for butter fat by the Babcock test, the payment is not on the same basis, and therefore justice is rot done between the two classes of pa rons. The Babcock test shows all the fat, while the oil-test churn shows only the churnable fat. The churnability of the fat depends upon a variety of conditions which at times make the results considerably inaccurate. This alone is sufficient reason for testing cream as well as milk by the Babcock method, both in separator and gathered cream factories. Besides securing the most accurate testing by the Babcock method it is more economical of time, as the composite sample plan may be used and the tests made twice or three times per month, instead of every time cream is brought to the factory, as it is generally done with the oil-test churn. Many of the factories running on the gathered cream plan still use the oil-test churn as a means of payment, the payment being made by the inches of cream and not by the amount of butter fat. Where the Babcock test has been introduced, it has generally given better satisfaction. The plan of using it as described by Winton and Ogden*, may be used with good results and is about as follows:

The cream gatherer is provided with a spring balance for weighing, a pail for holding the cream while weighing, small wide necked bottles with numbers on the corks corresponding to the patrons numbers, and with a sampling tube. The cream is emptied into the pail, weighed, and after being poured two or three times from ore vessel to another, a sample is taken and put in the bottle having the patrons number. When the driver returns to the creamery, the samples of cream in the bottles are emptied into a pint jar bearing the patron's number and containing some preservative. This gives a composite sample which may be tested as often as desired, at least once in two weeks.

Most creameries pay the same price per pound for butter fat in cream and in milk, when both are received at the same

*Conn. Experiment Station (New Haven) bulletins 108 and 119, Report 1894, pp. 214244. Also described in Farrington and Woll's "Testing Milk and Its Products."

factory. Some pay what is estimated to be the cost of separation more for fat in cream.

SAMPLING CREAM.

The composite sample system now generally used in creameries paying for milk by the Babcock test, may be applied to cream in the same manner. The taking samples of cream to be placed in the composite jar, or when taking a sample from the jar for testing, more care is required to get a fair sample than with milk. Cream being more viscid than milk, it is harder to get into a uniform mixture, while violent agitation causes air bubbles to be mixed in and on account of the viscosity they escape slowly. When each patron's cream is brought to the factory separately, the same methods of taking samples are adopted as are used for milk. It is generally best to have a small weigh can especially for cream, if any considerable quantity of it is received. As cream is subject to considerable variation in fat content, a sampling tube such as "the milk thief," with an opening slightly larger than is used for milk, is to be recommended as a means of taking the sample. In order that the cream adhering to the sampling tube from one sample may not influence the next, the tube should be filled and allowed to empty once or twice before the sample is finally taken. This is especially necessary when sampling milk after sampling cream.

Cream keeps about the same with the ordinary preservatives as milk under the same conditions. Some correspondents report more trouble in keeping cream when it is somewhat sour when sampled. It is hardly advisable to keep samples of cream over two weeks before testing. Where tests are made only monthly, as is done in some cases, a number of samples are liable to be partly spoiled, especially during warm weather, so a fair sample cannot be taken and consequently the results are unreliable and may lead to dissatisfaction.

Before taking a sample for testing from a composite sample jar, it is advisable to heat the contents to ninety or 100 degrees, especially when the cream is thick or has been in a cold place. This makes the cream less viscid and a uniform mixture more certain to be obtained after pouring carefully from one jar to another a few times. Raising the temperature this high does not lighten the cream enough to appreciably

affect the reading of the test. If the cream contains less than 30 per cent of fat, an ordinary 17.6 cubic centimeter pipette may be taken as when testing milk and the Winton* cream bottle used. From two-thirds to three-fourths of a 17.5 c. c. measure of acid is as a rule found sufficient for use in cream.

If a creamery receives cream from farm separators, it will be found most satisfactory to have the separators set to skim cream containing between 20 to 30 per cent of fat.

However, considerable of the separator cream tests more than this, so it cannot be read without some modifications.

DILUTION.

In the hands of a careful operator, cream may be tested with sufficient accuracy for most purposes by the use of milk bottles. A convenient way of doing this is to place a pipetteful of the cream to be tested in a cup, to this is added two pipettesful of soft or distilled water. If the cream is known to be very rich, so it will test over 30 per cent, more water should be added. The whole is then thoroughly mixed and a 17.6 cubic centimeter pipetteful of the mixture taken and tested as milk, except somewhat less acid is needed. The reading is multiplied by the number corresponding to the fractional part of the mixture occupied by the cream.

Where cream bottles are at hand and the cream to be tested can not be read on the scale, a mixture may be used composed of equal parts of cream and water.

A large number of tests made by this dilution method, showed it may be made as accurate as any that are based upon measuring the cream.

WEIGHING CREAM FOR TESTING.

When it is important that cream be tested very accurately, it is necessary to weigh out the amount rather than to measure it. The reasons for this are considered later in this article. For this purpose it is necessary to have a delicate balance and weights of the metric system.

The amount of milk made the basis of the Babcock test, and the amount of normal milk delivered by the 17.6 c. c. pipette is 18 grams. In weighing out cream, the amount taken may be made to suit the size of the bottle, and from the result the correct reading for 18 grams calculated. We have found the

This is the wide necked cream bottle reading to 30 per cent and is sold by all dealers in dairy supplies.

least number of errors are made and the most rapid work done by using cream bottles, and taking 9 grams for each test. The bottle may be put on one side of the scale and balanced with weights, or more conveniently with fine shot. When the bottle is balanced, 9 grams in weights are placed on the pan with the shot, and the well mixed sample of cream dropped from a pipette into the bottle until the balance shows 9 grams have been added.

Water is added to the cream to make up about 15 c. c. and about 14 c. c. of acid are used. When 9 grams are taken the

reading is multiplied by 2.

ERRORS IN TESTING CREAM BY MEASURE.

If cream is measured with the 17.6 c. c. pipette, as milk is, the result is too low for two reasons.

The 17.6 c. c. pipette gives an allowance of .1 c. c. for milk that adheres to the pipette. In measuring cream, on account of the more adhesive nature, a larger amount sticks to the glass. This amount in thick or sour cream is sufficient to change the results materially. The following figures show the effect of varying thicknesses of sweet and sour cream on the amount adhering to the pipette. The cream for the tests in the first column was measured out with a 17.6 c. c. pipette blowed out as when testing milk. The second column is the same cream measured with the pipette and the cream adhering to the pipette rinsed out with warm water and added to the measured cream.

From these figures it is evident no change in amount measured conld be made to remove this varying error. This error can be removed easily and completely by rinsing out the pipette with about one-third of its volume of warm water and adding this to the measured cream. When large quantities stick to the pipette, it may be loosened by placing a finger over each end after adding the water, and shaking violently.

All the results given in this bulletin are based upon this plan of rinsing out the pipette after measuring the cream.

THE ERROR DUE TO THE LOW SPECIFIC GRAVITY OF CREAM.

The size of this error varies with the per cent of fat. The 17.6 c. c. pipette delivers 18 grams of milk, but less of cream, as it is lighter, on account of the larger amount of fat present This error can be removed by weighing the cream, but for factory use there is considerable objection to this method, especially where the amount of cream received is small. Very few factories are provided with the necessary delicate scales, or with operators able to give the time and practice required for accurate weighing.

Another way of avoiding this error due to the low specific gravity of cream has been proposed in the form of a table showing the amount of the error for various per cents of fat. After testing a sample of cream, measuring it with the ordinary 17.6 c. c. pipette, the table could be consulted and the amount necessary to be added to remove the error due to the low specific gravity could be ascertained. As it is easier to apply

this table of corrections than to weigh the cream for testing, it was decided to make a series of experiments to determine if such a table could be adopted for use where the composite sample system is used, and be accurate within the limits reasonable for practical use. In doing this, the specific gravity of cream from 10 to 50 per cent was ascertained at intervals of 5 per cent as described in particular under a separate heading elsewhere. The conditions were made, as near as possible, such as would be present in testing composite samples. The error that would be made by measuring cream of these different per cents of fat was calculated by proportion. For example, the specific gravity of cream, containing 30 per cent of fat, was found to be .939. We then make a proportion as follows: