ancy was very large. Yet a careful renewed search thru the microscopic preparations failed to show any such large numbers of bacteria as were indicated by the agar plates. The most natural explanation of these irregularities is that extra colonies developed on the plates from bacteria which got into them from outside sources, and this was taken to be the case until it was noticed that all but 32 of the 175 counts were made on milk samples containing less than 30,000 individual bacteria per cubic centimeter. The localization of the majority of these counts among the milk samples containing few bacteria indicates that the majority of the counts of this type were caused thru the mathematical impossibility of getting a fair average when only a small number of objects are counted, from which to estimate a very

>> smear on the slide; since the field viewed under the microscope is exceedingly small and it would take altogether too much time to pass all of the thousands of fields in each smear under the microscope. Accordingly, we examine only one hundred of these fields, count the number of bacteria found, and multiply this number

by the proper factor to give the ap- GROUPS OF BACTERIA IN POOR QUALITY proximate number per cubic centimeter. If the milk contains only a very few bacteria, we may not even find one in one hundred fields of the microscope, and so use too small a number in the computation. On the plates a much larger fraction of a cubic centimeter of milk is examined in those milks which contain few bacteria than in the case of the microscopic slides. In making comparisons of counts from low-count milk samples, the probability is that the microscopic count will be lower than it should be. This is quite probably what happened in the 143 cases in which the colony count on the agar was greater than the individual count on the slides. Many of these discrepancies were not large, and under these conditions where so few bacteria were present the microscope makes no mistake in placing it in the "good" or Grade A class even tho the number of bacteria found is not quite as large as the number found by the plating method. No difficulty of this sort arises in milks containing a sufficiently large number of bacteria to be found readily under the microscope. In the case of high-count milk, the

advantage of examining a relatively large amount of milk is held by the microscopic and not by the plate method. It is possible that some bacteria are so small, or stain so lightly, that they are commonly overlooked under the microscope altho they would form colonies on the agar plates. There are few indications, however, of such instances, and it is believed that they are rare. The effect of such varieties if present would be to make the microscopic counts lower than they should be.

Let us return for a moment to the hypothetical Comparison methods of counting seeds. The results of the studies with counts on the counts of bacteria show that the commonly of seeds. used agar plate counts are much too low because of an error similar to the one caused by counting the beet seed balls as a single seed. They also show that there is an error of smaller size due to the presence of living bacteria which do not grow; an error similar to the one caused by the failure of some of the seeds to grow. The indications are also that where the work of making bacteria counts is carefully done, errors due to contaminations, while they do occur, are much less important than would be the similar error caused by weed seeds present in soil.

None of these errors affect the microscopic counts, the errors in this case being largely those which come from the difficulty involved in making accurate counts of objects so tiny that they can not be seen except with high magnification under the microscope. The claim which is frequently urged against making counts of bacteria microscopically based on the fact that dead bacteria cannot be readily distinguished from living ones does not prove to be of any significance in the case of fresh unpasteurized milk; because of an apparent absence or practical absence of dead bacteria in such milk. There are circumstances, moreover, in which a total count of the bacteria is of greater importance than a count of the living bacteria; and in fresh milks the dead bacteria give just as much information concerning the past history of the milk as do the living ones. Under certain special circumstances where preservatives have fraudulently been added to the milk, it is especially desirable to be able to detect the dead bacteria; a thing which cannot be done by means of the agar plating method.

In view of the manifest impossibility of so perfecting Conclusions. either the agar plate method or the microscopic method that absolutely accurate counts of bacteria can be made; or even of controlling the errors which have been discussed, it is fortunate that comparisons of real value can be made without counts more accurate than they now are when properly made. There is no necessity for us to be able to count the bacteria in milk so accurately as to tell the difference between a sample containing 5,657 bacteria and one containing 6,756 bacteria. In looking at a distant hillside, it is not possible to count the trees thereon. Never

theless, it is ordinarily quite possible to form an accurate judgment as to the area covered by woods, by wooded pasture, and by meadows. Such a judgment may have as great a value for the purpose in hand as would one based upon an accurate count of the number of trees per acre of land. The fact that accurate counts of the number of bacteria in milk cannot be made does not prevent the use of either the agar plate method or the microscopic method as a means of distinguishing with a satisfactory degree of accuracy between milks containing few bacteria, a medium number, and a large number. The fact that this is all that is necessary for commercial and control work has been shown in the inspection work which has been done for the City of Geneva during the past three years, a description of which will be given in a forthcoming bulletin.

It is evident from all of this work that popular ideas of the number of bacteria in milk, based as they are upon the counts made by the plating method, must be revised. The results obtained by the microscopic method of counting bacteria in milk show these to be much too low.

A

CONTROLLING A RADISH PEST. *

F. H. HALL.

A serious pest with which the grower of truck crops must contend is the cabbage maggot, or radish triple-action maggot. This little grub, the larva of a fly much pest. like our common household nuisance, often ruins the prospect for a good crop of early cabbages; it so tunnels, rasps and weakens the roots of plants of late cabbage in the seed-beds that the grower must frequently send away for his plants, and it all too commonly mutilates and deforms midseason radishes so that only a small portion of the crop is fit for the table.

The work of the insect as a cabbage pest has been discussed in a previous bulletin of the Station (No. 419); but its activities on the radish, tho of less economic importance, are undoubtedly better known to the home gardener. Hardly a season passes that the tunnels of the little white grubs are not found so abundant in many radishes that these must be thrown away. Occasionally the numbers of the pest have been so great that it has been practically impossible to get a perfect radish in early or mid-June.

Time of
injury.

The radish is a spring and summer crop, and the maggot flies appear at just the right time, from the second to the last week in May, so that their eggs are laid and the little grubs hatch and become active at their destructive work just when radishes in most home and farm gardens are making their best growth. Radishes that reach pulling size before or during the first week in June, in normal seasons, usually escape injury from the maggots; but to have the roots ready by this time the seed must be sown in March or early April. This is much earlier than most gardens are ready for planting. Such early sowing of radish seed is necessary to have the roots mature in late May or early June, since the radish does not grow very fast. until the soil warms up somewhat. In tests extending over four years, it was found that it required two months or more to grow

*

Reprint of Popular Edition of Bulletin No. 442, November, 1917; for Bulletin see p. 460. [Plates XXXI and XXXIII illustrated this text.]

radishes from seed sown during the last half of March, about seven weeks to grow them from seed sown in early April, six weeks from seed sown in late April and only four or five weeks from seed sown in May. The maggots disappear soon after the middle of June, but radishes maturing after this time, except under very favorable. conditions, are liable to be tough and too pungent for satisfactory To grow good radishes without protection, then, it is necessary to prepare the ground and sow the seed just as early as weather conditions and the physical condition of the soil will admit. But investigations made at the Station prove that the remedy that works so well for the grower of cabbage plants for the late crop is also valuable for the radish grower. This is, to cover the beds

use.

Protection
possible.

with cheesecloth.
During all four years of the tests, 1914-1917, both the radish
maggot and flea beetles were well controlled by the use of such
cheesecloth screens; and in three of the years the yields of radishes.
were much increased by this method. In 1917, owing to the wet,
cold spring, the number of maggots was much reduced. Injuries to
radishes, whether screened or unscreened, were comparatively
slight, and the increase in yields under screening was much less
than in previous years. The summarized data for the four years
are shown below:

TABLE I. SUMMARIZED YIELDS OF SCREENED AND UNSCREENED RADISH

PLATS FROM 1914 TO 1917.

(Counts and weighings are based on rows of one hundred feet in length.)