Atkins — Errors in Hydrogen Ion Determinations of Soils. 343 



of them will settle out sufficiently to give a portion which is satisfactory for 

 use. . This, however, may involve a new source of error, the biological production 

 of carbon dioxide, especially in the more fertile soils. This is quite insufficient 

 to aft'ect appreciably the turbid extract, which is heavily buffered by the soil 

 particles, but may lead to the lowering of the pH value of the supernatant 

 liquid, especially if alkaline. Errors of from pH 0-1 to %>H 05 may be 

 encountered from this source; the agitation of the liquid, with several changes 

 of air, serves to extract this carbonic acid, and a result approximating to the 

 true initial value may be obtained. Thus the method of subsidence, though 

 convenient where practicable, has its limitations. 



As a routine practice in all his work, the writer has cleared soil solutions 

 by means of an electric centrifuge capable of running at 9,000 r.p.m., but 

 usually run at a lower speed owing to much trouble due to breakage of tubes. 

 This has proved quite satisfactory with almost all soils examined, though a 

 few — while clear enough to permit of print being read through the suspensions — 

 were yet slightly turbid. A number of soils give clear extracts with a slight 

 yellowish tint; with these a comparator should always be used, otherwise the 

 pH values are too low by 01, 0-2, or more, with indicators such as phenol red 

 and brom thj^mol blue, or too high with indicators, such as methyl red, in which 

 the red to yellow colour change is in the reverse direction as compared with 

 phenol red. Should a tube break in the centrifuge the liquid must be rejected, 

 even if clear, as it is markedly too alkaline owing to contact with the freshly 

 broken glass surface. The use of colloidal iron was tentatively suggested by 

 Gillespie (1920) as of possible value for clearing solutions. According to the 

 writer's expexience its use is liable to give an error in the direction of an 

 increase in the acidity. Gimingham (1923) has recently introduced percolation 

 through a column of soil as a methocl of 'obtaining a clear solution. 



The use of a centrifuge would not be possible everywhere, and filtration, 

 if peimissible, would be far more convenient. Olsen (1923) states that with 

 Swedish "Berzelius" acid-extracted filter-papers his colorimetric were within 

 pH 0-2 of his mifiltered electrometric determinations made upon wood and 

 meadow soils, using one volume of soil to one of water. These soils were mostly 

 rich in humus. 



With arable soils, however, he records errors of as much as pH 0?>. It 

 may be remarked that Olsen uses distilled water for his extractions, as is also the 

 writer's practice, but Wherry uses any natural water close to neutrality, which 

 appears to introduce an error owing to the buffer action of a natural water at 

 pHl-2, due largely to silicates .and bicarbonates. Wherry's (1922) methods 

 are admittedly approximate, being designed for rapid field work. Salisbury 

 (1922), too, follows Olsen in filtering his solutions made from 10 grams of 

 undried soil with 50 c.c. of water neutral to brom thymol blue. The first filtrate 

 is discarded, and the results obtained by this technique agreed, in the samples 

 tested by him, with those given by clearing with the centrifuge. Kelley (1923) 

 and others have tried diluting turbid extracts to two or three volumes, _ and 

 using a comparator. The error given is said to be small, and with highly 

 buffered clay soils this is probably true. The writer considers indiscriminate 

 dilution to be a risky procedure, though at times it proves useful. 



The tables which follow show that filtration may introduce serious errors. 

 Thus alkaline soils appear nearer neutrality when filtered through an acid- 

 extracted filter-paper, and acid soils may have their acidity shown as erroneously 

 low even when filtered through acid-extracted paper, and to a more marked 

 extent when ordinary filter-paper is used, such as Whatman, No. 1. Munktell's 

 No. is evidently the "Berzelius" paper used by Olsen, and is acid-extracted. 



