Arising from this work, it was thought that size of cation might influence the selectivity of other commonly used basic dyes. In the present work, fourteen basic aniline dyes have been investigated, ranging in molecular weight from 264 to 1,341 (Table 1). A variety of formalin- or Helly-fixed tissues has been used, including frog pancreas, liver, ovary, intestine, stomach and kidney, rat trachea, stomach, pancreas and intestine, and monkey pancreas. Table 1Dye and source Molecular weight Cationic weight Percentage solubility Alcohol Water Alcian blue SGN15Q c. 1,341 1,341 (?) 6-0 9-5(I.C.I.) Victoria blue 4R(National Aniline Co.) 520 485 20-0 3-0 Janus green B(National Aniline Co.) 483 448 1-0 5-0 Bismarck brown R(National Aniline Co.) j 461 388 3.0 1-5 Methyl green OO(Griibler) 458 387 3-0 8-0 Hofmann violet(Merck) 422 387 7-5 6-0 Crystal violet(National Aniline Co.) 408 373 8-75 9-0 Nile blue A(National Aniline Co.) 733 316 5-0 6-0 Safranin T (Merck) 351 316 3-5 4-5Brillantcresylblau (Griibler) 318 282 2-0 3.0Toluidine blue (Griibler) 306 271 1-75 3-25Pyronin (Griibler) 302 267 0-5 9-0 Azur A(National Aniline Co.) 292 257 ? ? Thionin (Grubler) 264 229 1-0 1-0(1) Molecular weights and solubilities in alcohol and water are from ref. 1. Cationic weights are calculated from data from the same source. (2) One molecule of nile blue A gives rise to two cations on ionization.In initial experiments 0-01 M aqueous solutions of the dyes were used, the stained, air-dried sections being mounted directly in 'D.P.XS (synthetic mount-ant, British Drug Houses). It was found that in these conditions sections were markedly overstained by certain dyes (notably victoria blue 4.R, bismarck brown R and hofmann violet). It is noteworthy that most of the dyes concerned are more soluble in alcohol than in water (Table 1), and it is possible that this is a reflexion of some property common to this group of dyes, which is responsible for a loose, non-ionic bonding with tissue components. This is to some extent confirmed by the fact that thorough dehydration in ethyl alcohol (one minute each in 90 per cent, 95 per cent and absolute alcohol) was sufficient to remove the excess dye. The results about to be described were obtained on sections stained for 15 min. in a 0-005 N solution of the dye in veronal-acetate buffer at pH. 4-25, washed in tap water, dehydrated in alcohol and mounted in 'D.P.X.\Under these conditions alcian blue (weight of cation about 1,341) stained nuclei minimally, and did not stain nucleoli or cytoplasmic ribonucleic acid. Victoria blue 4R, janus green, bismarck brown R, methyl green OO, hofmann violet and crystal violet (with cationic weights from 485 down to 373) stained nuclei more deeply than nucleoli or cytoplasmic ribonucleic acid, while safranin and nile blue (cationic weight 316) stained nuclei, nucleoli and cytoplasmic ribonucleic acid with approximately equal intensity. Brilliant cresyl blue, toluidine blue, pyronin, azur A and thionin (cationic weights from 282 down to 229) stained nucleoli and cytoplasmic chromidial substance more intensely than they stained nuclei. From these results it appears that dye cations having a combined atomic weight of between about 350 and 500 are able to penetrate into and become attached to nuclear deoxyribonucleic acid (DNA), but do not as readily stain the nucleoli and cytoplasmic granules of ribonucleic acid (RNA), which are presumably denser and better stained by smaller dye cations (of a combined atomic weight of less than about 280).Differential staining of DNA and RNA by mixtures of two basic dyes (as in the classical methyl green/ pyronin method) is now readily explicable. The proportions of the two dyes are chosen in such a way that the larger cation (methyl green) competes under favourable conditions for the DNA, while the smaller cation (pyronin) stains only UNA. If the principal factor in such differential staining is the weight of the dye cations, it should be possible to reproduce the 'methyl green/pyronin effect' (differential staining of DNA and RNA) with other pairs of basic dyes, provided that the dyes chosen are of sufficiently different cationic weights. The dyes must also, of course, be of contrasting colours, be mixed in appropriate proportions and be used at the correct pH. To confirm this prediction, 1 per cent solutions of several basic dyes were prepared. Paraffin sections were stained for 15 min. in a mixture of a pair of such types (combined in various proportions), rinsed in tap water, dehydrated thoroughly in alcohol and mounted in 1D.P.X.\ The following dye pairs stained DNA and RNA differentially under these conditions, when mixed in correct proportions: bismarck brown .R/toluidine blue, bismarck brown -R/azur A, victoria blue 4.B/hofmann violet, victoria blue 4R/ safranin, victoria blue 4R/pyronin, janus green/ pyronin and janus green/safranin. In every case, the dye with the larger cation stained the nuclei, while the smaller cation stained nucleoli and cytoplasmic chromidial substance. A few nuclei, especially the small nuclei of frog red blood cells, stained with the smaller cation instead of the larger. This may be due to such nuclei lying wholly within the plane of the section; in such a case the nuclear membrane will be intact, and prevent access of the larger cation8.Dye pairs in which there was a large difference between the weights of the cations gave particularly good discrimination between DNA and RNA. Thus victoria blue 4R/pyronin or janus green/ pyronin (mixed in a proportion of one part by weight of the larger cation to four parts of the smaller) gave results which compared favourably with those obtainable with the classical methyl green/pyronin method. Where the difference between the cationic weights of the two dyes was not large (for example, victoria blue 4R/hofmann violet), or where incorrect proportions of the two dyes were used, good differential staining of nuclei and nucleoli was sometimes obtained, but cytoplasmic RNA stained with both dyes. This is taken to indicate that the density of cytoplasmic RNA granules is intermediate between that of DNA and that of the nucleolus. The use of basic dyes of varying cationic weights can also give some indication of the change in density of a structure under different conditions. Thus the staining of so-called 'depolymerized' DNA with pyronin instead of methyl green in the methyl green pyronin method4 probably indicates that the heat, or other treatment to which the DNA has been subjected, has altered its structure in such a way as to increase its density, and render access of the larger cation more difficult.Dye-pairs in which there was relatively little difference between the cationic weights (for example, crystal violet/pyronin or safranin/thionin) gave little or no differential staining of DNA and RNA, irrespective of the proportions in which the dyes were mixed.It appears from these results that the chemical explanations which have been advanced to explain the specificity of the methyl green/pyronin method4 are unnecessary, the chief factor being the weights of the dye cations. The same factor explains the effect of the dye mixtures, malachite green/acridine red, methyl green/toluidine blue and celestine blue/ pyronin, which give the same empirical results as methyl green/pyronin4; malachite green has a cationic weight of 328, celestine blue 329 and acridine red 240. It should not be concluded from fche above that the weight of the cation is the only relevant factor in staining with basic dyes. Mordants, pJL, solubilities in various media, etc., may also play an important part in some procedures5.