dummy-one that does not vary, and the writer therefore thinks it better still to use a weighed sulphuric acid bulb or some other additional drying apparatus after the prolong to secure this uniformity until convinced by many such trials that (in the operator's manner of procedure) no moisture escapes absorption by the prolong. For the writer, moisture does escape from the prolong, in the dry as well as in the wet method of combustion, varying in the dry method from 0.000 per cent. to 0.08 per cent. in terms of carbon, according to the freshness of the calcium chloride, its state of division, and the speed of aspiration. In using calcium chloride Dr. Dudley directs that it be freshly dried before use. This is such a necessary precaution as to bear much iteration.

A calcium chloride tube filled directly from a half-emptied bottle and therefore without previous drying lost by aspirating dry air through for two hours, equivalent to carbon: 0.047 per cent.; repeated, lost o. 14 per cent.; again, gained 0.02 per cent. ; again, lost o.19 per cent.; again, lost 0.48 per cent.

A wash-bottle with water was followed by this calcium chloride, and that in its turn by a weighed sulphuric acid bulb. The bulb gained in weight 0.32 per cent. and 0.29 per cent., so that the drying of the calcium chloride previous to use is an absolutely necessary precaution.

It seems to be necessary with some potash solutions to aspirate air through before using. Several blank tests made without doing this showed a loss of weight, in the one case equal to 0.048 per cent. carbon and in the other o.019 per cent. carbon.

In the writer's experience a most serious difficulty with the wet method of combustion is, that occasionally results are obtained from 0.05 to 0.07 per cent. above the truth, due apparently to some chloro-chromic compound other than that absorbable by Prof. Langley's "pyro" mixture. At least the writer can offer no other explanation, the purifying train including: water kept cold for hydrochloric acid according to Porter Shimer; ferrous sulphate for chlorine according to Dr. Dudley; "pyro" mixture for chlorine, oxides of chlorine, and chlorochromic anhydride according to Prof. Langley; and silver sulphate in strong sulphuric acid for hydrochloric acid.

It is known that the wet method is unsuited for graphite de

terminations owing to a portion of the graphite being oxidized merely to carbon monoxide.

[CONTRIBUTIONS FROM THE CHEMICAL LABORATORY OF THE U. S. DepartMENT OF AGRICULTURE, No. 30.]

COMPARISON OF THE STANDARD METHODS FOR THE ESTIMATION OF STARCH.'

BY H. W. WILEY AND W. H. KRUG.
Received February 15, 1898.

HE study of the nature of the carbohydrate bodies which

Texist in foods has been vigorously prosecuted by chemists

in different parts of the world during the past few years. result of these studies has been to arouse new interest in the standard methods of starch determination. The numbers which are secured by different workers for starches vary so greatly as to lead to the belief that the estimation of starch is not such an easy matter as is generally supposed; in fact, long experience has convinced us that there are few operations in agricultural analytical chemistry which require greater skill, or are attended with greater difficulties than the estimation of starch in the presence of the other carbohydrate bodies.

At a recent meeting of chemists in England, it was stated in one of the papers that the methods of starch determination are now so easy of application and so accurate that there is no longer any excuse for being ignorant of the exact content of starch in any body. Stone, in Bulletin No. 34 of the Office of Experiment Stations, described a method which was easily seen to be erroneous, yielding only about half as much starch as was really contained in the samples examined. A cursory glance at the methods of analysis employed by him convinced us that the fault was insufficient saccharification, due to the weakness of the acid and the shortness of the time of heating.

In a subsequent article Stone called attention to the error of his previous statements, and proposed a modification of his first published method, which, as is seen below, gives very much better results. Having to estimate, annually, the starches in many hundreds of substances, we were led to make a compara

1 Read before the American Chemical Society and Section C of the American Association for the Advancement of Science, at Detroit, August 12, 1897.

2 This Journal, 19, 347.

tive study of the standard methods which have been used for starch determination, and of some modifications which seemed. to us worthy of trial. The standard methods which we employed were those described in "Principles and Practice of Agricultural Analysis," 3, 196-222, and 300-302. The materials which we used in our investigations were starches of commerce, the ordinary cereal grains, and a sample of the wheat which was the subject of the investigations of Stone, and which was kindly furnished us by Professor Stone. We give below a brief résumé of the work we have done.

The sample of spring wheat No. 1, which we received from Professor Stone, was examined for all of its principal constituents, according to the official and standard methods of analysis, and the data obtained by him and by us are given below. The moisture stated in the second column of the appended table was determined by drying for five hours in a high vacuum at 100°.

Before determining the starch it seemed of interest to take the sample in the state in which it was sent, as we assumed that it had thus been analyzed by Professor Stone. It also seemed to us that the digestion with malt extract prescribed by him was not long enough to insure the complete conversion of the starch into maltose and dextrins. To determine this, one set of samples was digested one hour at 60°, and another set two and onehalf hours at 50°, since Lintner states that diastase is most active at this latter temperature. The results obtained show that our surmise was correct.

The residues from the diastase digestion were all thoroughly washed with hot water and then examined with iodine under the microscope. In every case a large number of cells was found. which contained undigested starch, showing that the sample had not been ground to a sufficient degree of fineness. This is, therefore, another source of error in Professor Stone's work. The sample was then reground and the starch determined by three methods; viz., first, the Reinke lactic acid method, second, a method which differs from this by the use of five-tenths gram of salicylic acid instead of lactic acid, and third, the diastase method. The residues were again examined and in every case found free from starch, showing that the conversion had been complete.

The following results were obtained:

As is subsequently shown, the result obtained by the last method is more probably correct. The number for starch thus obtained, added to our per cents. of the other constituents gives us a total of 99.28. The separate determination of the sucrose, galactan, invert sugar, and dextrin, originally present, was omitted, and these bodies are included with the starch.

This work led us to undertake an investigation of other methods for the determination of starch which have been proposed from time to time, our desire being to avoid, if possible, the use of diastase and to shorten the time of procedure.

It seemed that this might be accomplished by polarizing the

products of hydrolysis and thus avoiding the labor incident to the gravimetric method of determining copper. To this end, several methods were tried. These methods depend either on the complete conversion of the starch into dextrose or on the formation of soluble starch, maltose, and dextrins.

THE GUICHARD METHOD.

Guichard' proposes to boil four grams of the material for one hour with 100 cc. of ten per cent. nitric acid (ten cc. strong acid (sp. gr. 1.42) to ninety cc. water) in a flask provided with a return condenser. The solution is cooled, filtered, and polarized. The starch is found by the formula

where a

and A

rotation in angular degrees, v = volume of liquid, starch transformed.

As our polarizations were made

with a Schmidt and Haensch instrument, the dextrose was determined by the formula

Three grams of substance are heated in a flask with eighty to ninety cc. of water until pasty, and the starch rendered soluble by heating from three to five hours in an autoclave, at from two to three atmospheres. The volume is completed to 100 cc. and the solution polarized. The gyrodynat of soluble starch prepared in this way varies from (a), 196.5° to 197°.

=

THE BAUDRY METHOD.

To 2.883 grams of the substance (for Schmidt and Haensch instrument) add from eighty to ninety cc. of water and fivetenths gram salicylic acid. Boil twenty minutes, dilute to about 190 cc., add one cc. ammonia, cool, dilute to 200 cc., filter, and polarize in a 400 mm. tube. Multiply the result by three. Baudry determined the gyrodynat of soluble starch to be (a) =

D

1 This and the following polarimetric methods are described in detail in Vol. III of "Principles and Practice of Agricultural Analysis."