475

ART. LX.-On the RHbz"cliwn and PotassiuJn T1'ilwlides / by H. L. WELLS and H. L. W HEELER. TVitl~ their Cl'Ys­ tallogNtplly / by S. L. PENFIELD.

THE discovery of a series of cmsium trihalides* has led the writers to investigate the analogous rubidium and potassium compounds. The following table gives a list of the bodies which we have been able to prepare, together with the cmsiulll series fot· comparison. The compound KI . I. had been previ­ ously prepared by J ohnson.t * On a Series of C::esium Trihalidcs, by H. L. Wells; including their Crystal­ lography, by S. L. Penfield. This JO\1l'ual; III, xliii, 17. t J .. Chem. Soc., 1877, 249. 476 Well8, Wheeler and Penfield-Rubidium and

CsI. I. RbI. I. KI. I. Ci3Br. I. CsBr. BrI RbBr. BI'I KEr. BrI CsC!. ErI RbC]. BrI CsC]. CII RbC]. ClI {(C]. CII CsBr. Er. RbBr. Br. CsCl. Br. RLCI. BI', CsCl. CIBr RbC]. C]Sr It is to be noticed that there is but one member lacking in the rubidium series to make it as complete as that of cresium. We have repeatedly tried to pr~pare this compound, RbBr. I., using alcoholic solutions of varying strength and great con­ centration at low temperatures, but with no success. The failure to make this body doubtless depends upon the com­ parative instability of the rubidium series. We have even attempted to prepare RbOl. I. and RbOI. 01., corresponding to which no cresium compounds could be made, but, as was anticipated, these efforts were entirely without success. In the potassium series only those bodies could be prepared which correspond to the more stable cresium and rubidium compounds. They show a great decrease in stability in comparison with the rubidium compounds. A produet was obtained at a very low temperature, which was probably KBr . Br., but we did not make a satisfactory analysis of it. We have attempted to prepare a number of sodium and lithium trihalides. There is no doubt that some of them exist, but they are so extremely soluble aud unstable that we have abandoned work in this direction. JJfetlwd if Preparation. - The rubidium and potassium compounds are made, like the cmsium series, by dissolving a Normal halide with the propel' halogen or halogens in water with the aid of heat and cooling to crystallization. The members of the rubidium series, being very soluble, require very concentrated solutions for their preparation. The potas­ sium compounds, being still more soluble, require the greatest possible degree of concentration and are usually best obtained by exposing the solutions for a considerable time to a winter temperature, evaporation in the desiccator being sometimes also necessary. Oolor.-The colors of the rubidinm and potassium com­ pounds are very similar to those of the corresponding memo bel'S of the cresium series, but since they usually form larger crystals their apparent color is generally somewhat darker. They vary in color from brilliant black in RbI. I. and KI. I., through various shades of yellowish-red and orange to bright yellow in the compound RbOl. OIBI'. In all the compounds Pota88i'M1n T-l'ihalide8, W'tt!l, tlteir OI'Y8tctllogmph,y. 477

that have been prepared the color becomes lighter as the sum of the atomic weights of the three halogen atoms decreases. Stabil-ity.-It has been fonnd by experiment that the potas­ sium trihalides are much less stable on exposure to the air than the corresponding' l:ubidium compounds, while these in turn are less stable than the members of the cresinm series. The same relative stability of the three series is shown by the temperatures at which they are completely decomposed by rapid heating as given below: App1'oximate temlJe1'atllre oj' whitening. OsI, I. 3:30 0 RbI, I. 270 0 KI. I. 225 0 OsBr, Bd 320 0 UbE1', BrI 265 0 KB1', BrI 180 0 OsOl, Oil 290 0 RbOl, Oil 265 0 K01, Oll 215 0 CsCI. Bd 290 0 RbCl, BrI 200 0 ______CsBl' , BI'. ]600 RbBl', Hr. 1400 ______OsOl, CLBr 1500 RbCl. ClBt, 110 0 ______CsCI. 131'. 1500 RbCl , Br., 80 0 • __ • ___ _ F1t8ibility,-The melting-points of the analogous com pounds become lower from cresium to potassiuUl. In the open capil­ lary tube RbI, I. melts at 1940 and RbOl . OIl at ~08°, while all the other rubidium compounds whiten without melting, The potassium compounds give practically the same melting­ points in open as in sealed tubes, The following table gives the appl'oximate melting-points in sealed tubes:

CsI , I. 201 °-20So RbI, I. 190 0 Kl. I. 3So* CsBt', Bd 24;i°-24So HbBI'. BrI 225 0 KBr, 131'160° CsCl, Oil 225°-2300 RbOI. Oil 180°-200° KOl, Oil 60° CsCI. 131'1 225 0 -2350 RbOI. Hr1205° ______OsBt, ,131'. 180 0 I{bBl'. 131'., whitens ______OsCl , CiBr 205 0 nbCl , OlI~r whitens ______OsOI,Br,191° RbCl.Br.76°? ------

Be/i,aviol' witl~ Solvent8,-The extreme solubility of the rllbidium and potassiulll tdhalides in watel' has already been refel'l'ed to, and it has been pointed ont that the members of the potassium series are the most soluble, The rubidium compounds which contain iodine can be recrystallized fl'om water without difficulty. These fOUl' bodies coutaining I'U­ bidinm and iodine al'e sufficiently stable to be soluble in alco­ hol, while the remaining rubidium compounds, as well as all the potassium componnds, are more or less readily decomposed by alcohol with the separation of normal halides. Ethel' decomposes all the rubidium and potassium compounds, leav­ ing normal halides undissolved, * Johnson gives 45° for the molting-point of this compound (\. c.) 47R Well8, lVlLeeler c£nd Penjield-Rubidittm. and

Crystallography, The rubidium trihalides crystallize in the orthorhombic system and are isomorphous with the corresponding crnsium compounds, showing a close similarity both in crystalline habit and in axial ratios, . The forms which have been observed are a, 100, i-'i cl, 011, 1-~ b; 010, i-~ j; 021, H c, 001, 0 e, 102, i-i m, 110, I p, Ill, 1. With the exception of the pyramid p, which was observed as a small face only on RbI, I., these are the same as were observed on the cresium tdhalides, while the brachydome g, 012,t~ which was found only on OsI , I21 was not observed on any of the rubidium compounds, Of the three potassium trihalides, which were examined, only one, KBr, RrI, was orthorhombic, like the cmsium and rl'tbidium compounds, The others, KI, T. and KOI, 01I, arc monoclinic, but they can be referred to axes which are similar to those of the orthorhom hic series, The cleavage of the rubidium trihalides is perfect parallel to c, less perfect parallel to It; neither is easily produced, The crystals are very brittle and usually bt'eak with a conchoidal fracture, The potassium trihalides are exceedingly brittle and no cleavage was observed, The optical properties were not studied, owing to the difficulty of preparing orientated sections, In the following table the axial ratios of all of the alkali­ metal trihalides are given, arranged as in the crnsium paper, In the table of angles, those which were chosen as fundamen­ tal are marked by an asterisk, r. Series with iodine, If. r-----"----.., ,-'---"----.., U : b : c a : b : c CsI, I. O'li824: 1: 1'1051 1 : I '·W55 : 1'6 19tj ~ RbI, I. 0'6858: 1·: 1'1234 l: 1'4582: I'Gatll Kl,l. J 0'7065 : 1: -- -- I : 1'4][,4 ! Monoclinic, a = 86° 47f a = 86° 47'.!-' CsBr, I. O'6U16 : 1 : 1'14 HJ 1 : 1'4460 : 1'6511 CsBr, Bd 0'7203: 1: 1'1667 1 : 1'3882: ] '6196 { RbBl',Brl 0'7130:1:1'1640 1 : 1'4025 : l'(i325 KBl', BrIO'7l58 : 1 : \'1691 1 : 1'3970 : 1'6333 . j OsCl, Bd 0'7230: 1 : 1'1760 1 : 1'3831 : 1'6268 ! RbCl. Bd 0'7271: \ : 1'1745 1: 1'3753: 1'6153 CsOl, CII 0'7873: 1 : 1'1920 1: ]'3563: 1'6167 { RbCI, CII 0'7341 : 1 : 1'1963 I : l'a622: 1'6296 KCl, CII {O'7835 : 1 : 1'2204 1 : 1 '3633 : 1 '0638 MOlloclinic, a = 83° 20' a= 83° 20' Potassi'nm. T1,i/lCtlides, 'I.()itll. tltei1, Ol'ystailogmpll.y. 479

Series without iodine. \ CsBl'. BI', 0'6873: 1: 1'0581 1 : 1'4550 : 1'5395 -I RbBr. Br. 0'6952: 1 : 1'1130 1 : 1'4384: 1'6023 5 CsCI. Br. 0'699: 1: -- -- 1 : 1'430 : - - -- t H,bCI.Br. 0'70 : 1: 1'1260 1 : 1'43 : 1 '61 1: 1'3917: l'E638 { CsCI . ClBr 0'71 86 : 1 : 1'1237 RbC]. ClBr 0'7146: 1: 1'1430 1 : 1'3994 : 1'5995

m"m,llO,,110 d"d, OIl "OIl 8,., e, 102" 102 RhI. I. *680 53' *96° 39' 78° 38: KI.I. *70 34 RhBr.BrI 70 58 *98 40 *78 27 KBr. BrI 71 ]2 *98 55 78 28 RhCI.BrI 72 2 *99 lot *77 51 RbCI. ClI 72 34 *100 13 *78 21 KCl.ClI 72 54 *79 8 RoBr. Br" *69 37 *96 10 77 24 RhCI. Br; *70 approx. *96 58 76 approx. RbCl. ClBr 71 6 *97 38 *77 18

A comparison of the axial ratios of the trihalides shows that the replacement of cresium by rubidium, and in one case by potassium, has little 01' no effect on the form, while in two of the compounds potassium causes a change in symmetry with­ out mnch .change iu the axes. It is evident that the rubidium salts like those of cresium may be arranged in two symmetrical series, one with and the other without iodine, in which the ratio of two axes remains nearly constant throughout while the third varies, and the conclusions which were arrived at in our previous paper concerning the constitution of the cresium trihalides, are confirmed by t4e rubidium compounds. The rubidium trihalides have a strong tendency to crystal­ lize and the solubility is such that, from solutions of not over 50 c.c. in volume, large and magnificent crystals, several centi­ meters in length, can readily be obtained. The size of the crystals seems often dependent only upon the volume of the solution and the size of the vessel containing it. Many of the large crystals are complex, being built up of smaller ones in parallel position. Some of the crystallizations were as beauti­ ful as any that we have ever seen. The rilbidilllll trihalides containing iodine were measured. at ordinary temperatures; those without iodine and. the potassium trihalides at about 0° O. It was found that the stability of the compounds increas_ed very rapidly with a diminution in tem­ perature and, hy working in the cold, no difficulty was experi­ enced in making accurate measurements of the more unstable E'alts. It is not considered necessary to give with each trihalide 480 lI'e1l8, TVllee1er (fJul Pe?lf/eld-Rubidi'll1n and

a tahle of. measured and calculated angles, but in all cases, where a sm'les of accurate llleaSlll'elllents were obtained, they agreed closely with the calculated.

I. 2. :1.

RbI. I,. The forms b, 0, 1)/" d, .I; e and p were observed. Of thesef and p were always small and frequently wanting. The ~labit is shown in fig. 1. .. RbBr. BrIo The forms of a, 0, 111" cl and e were observed. The habit is shown in fig. 2. RuOl. Bl'I. The forms rt, d and e were observed. The pillacoid a is nsually wanting and tIle simple habit shown in fig. 3, prevails. RbOl. OlIo The forms a" d and e were obsel'ved. The habit is shown in fig. 4.

RbB)' ,Br2 • The forms a, b, ln, d and .f were observed. The habit is shown ill fig. 5. RbOZ, ]]1'2' The forms b, 0, 111, d and e WCl'e observed. The habit is show1l in fig. G. The tendency of this salt is to crystal­ lize in small scales; it is also the most unstable uf the rubidium

4. 5. 6.

series, so that we considered it fortunate that we were able to make out the axial ratio. The only faces which yielded good reflections were 7) and d, which established accurately the rela­ tion betweeJl the h and I: axe~, while only approximate meaSllre­ ments w()re obtained from the other faces. RbCt. OlBr, The forms a" 0, In, d alld e were observed. The habit is like fig. 2, except that 0 is wanting, KI. L This occurs ill very simple monoclinic crystals. If the solution is cooled slowly it forms in stout prisms, but by rapid coolina' a net work of fine needles is obtained. In order to make thi~ salt and the monoclinic KOI. OIl conform to the position which has been adopted for the orthorhombic trihal· ides it is necessary to deviate from the ordinary custom and make the clino-axis slope from right to left instead of from back to front. The faces are ta.ken as b, 010, i-i; 0, 001, 0 and m, 110, I. The crystals are not sufficiently modified to deter­ mine more than two axes, but taking as fundamental measnre­ ments, bAm, 010 A110 = 54° 43' and CAO (re£intrant angle of twin crystal)= 6° ~5' the following axial ratio is obtained a: b ='70li5: 1; a=010AOOl = 86° 47t'. The angle mAC, 110 '" 001, was measured 91° 55' and 91° 50', calculated 91 ° 51'. Fig. 7 represents a twin crystal in the above position. Fi~. S represents a simple crystal in the ordinary monoclinic positlOn,

:-I. 7. R. c

6 !!! m '" In

,;it!l rt as the clino axis. The axia.l ratio for this position is, a:b = 1'4154:1; (d= 86° 47t'. KBl'. BrI. The forms {f" b, n, it, f and e were observed. The habit is shown in fig. 9. This salt differs from all of the otEer alkali-metal trihalides in having tIle brachy prism n, 120, i.2, instead of the nnit prism m. The fundamental measure­ ments were aAn 100 A 120 = 55° 4' and dAd. 011 A 011 = 98° 55'. . , KOl. UlI. This crystallizes in long needles belonging to· the monoclinic system, fig. 10. Taking b as the clino axis the 10. forms are a, 100, i-~ " b, 010, i-i,. c, 001, 0,. w, 032, i-i, and e, 102, t-1. i,x.~ The measnrements taken as funda- ~ _.f; mental are C!.",b, 001.,,010 = 96° 40', e eAe, 102 A 102 = 79° 8', and CAW, 001" 032 = 66° 35' from which the following axial ratio was calcnlated, a: b : c= '7335 : 1 : 1'2204, 482 Well8, Wh,eeleT and Pe?~fleld-R1lU£(UU?n and

a = 83° 20.'. If taken in the ordillary m'onoclinic position with e as the prIsm 110 and w as the OI'thodome 101, the axial ratio from the above measurements becomes d: b: C= '8319 '1 . '4544 f3 = 83° 20'. .. . lIIetlwd of Analysis. ~h~ met~lOd~ nsed for ths analyses of the potassium and rubIdIUm trlhalIdes were exactly the same as those mentioned in the article on cresium trihalicies. . The crystals were prepared for analysis by pressinO' between papers and at the same time crushing them some~hat. In s0!I1e cases, where the bodies were vel'y ensily decomposed, thIS was done in cold weather out of doors, but even with this precaution it was not possible to dry them very thoroughly 01' to avoid a considerable amount of decomposition.

RbI. I,. This body can he prepared by dissolving 55 g. of rubidium iodide in enough water to make a solution of 50 c.c., adding 60 g. of iodine, warming nntil solution takes pInce and cooling to ordinary temperature. A mass of large crystals in parallel position, forming steps, is usually formed. Ualculated Analysis gave for RbI. T•• Rubidium ______18'32 18'~2 lS'33 Iodine ___ . ______81'07 8l'67 A specific gravity determination, made in the mother-liquor at 22° gave the number 4'03. This cannot be considered very exact on account of the difficulty of obtaining the mother-liquor in such a condition that it neither dissolves nor deposits the substance. A sample of mother-liquOI', of specific gravity 2'19, was found to contain 1'61 g. of RbI. I. in 1 c.c. The com­ pound therefore dissolves in about one third its weight of water at 22°. It is interesting to notice here that under nearly the same conditions, the corresponding cresium compound, OsI . I., requires more than one hundred parts of water to dis­ solve it. It is expected that tllis great difference in solubility will form the basis of a useful method for separating the two metals. RbBr. BrIo This componnd can be readily made by dissolving, with the aid of heat, 30 g. of iodine and 20 g. of bromine in a saturated aqueous solution of 40 g. of rubidinm bromide and cooling. The facility with which this body crystalli7.es is remarkable. The large crystals have a color and luster much like the mineral pyrargyl'ite, "ruby-silver." PotasSi1t1TL TrilwUdes, witll' thei1' Crystallography. 483

Calculated AnalYHiH gave for RuBr . Br 1. RubidillllL _ _ :J2'7fl 22'95 Bromine ______45'ln 42'95 Iodine ______31'11 34'10 An approximate specific gravity determination, made with the mother-liquor, gavc the number H·S-I,. An analysis of the mother-liquor showed that it contained about ++ per cent of RbBr. Bd. The lllothet'-liquol' of the corresponding cmsium compound contained only 4'4,5 per ccnt of O::dk. Br1.

RbOl.lJrL This body can be made by adding 27 g. of bromine and +2 g. of iodine to a saturated aqueous solution of 40 g. of rubidium chloride, warming until all is in solution and cooling. It forms magnificent crystals which can be readily recrystallized from water. Unlike the corresponding cmsiull! COll! pound, it does not change its com position by recrystallization, hence it is probable that it is a true chemical compound and not a mixtnre of thc isolllorphou::; bodies RbBr. 13rI and RbOI . Oll. Analysis gnve, O)'iginal lith J'ccrys- Cnlcnlated crystals. tnllization. for RbCI . Br 1. HubiJillllJ ______2u'u7 2,'34 2u'Uu Chlorine . ______I (l"(l5 10'8:J Bromine _____ . _ :!4'8() 24'3n Iodille ______3t;'13 38'72 RbOl.OlL A convenient method for prepa"ing this compound is to pass chlorine into a warm, concentrated solution of rubidium chloride, containing the calclllated amount of iodine, until the iodine is just dissolved. If too Uluch chlorinc is used, the compound RbOl) is formed, which we shall describe in a futUl'e article. It is best to stop adding cltlorine while the solution is still colol'cd red by ioLline. On cooling the liquid the compound separatcs, usually in large fiat gl'OUpS of parallel crystals. Cnlellinted for AllnlYHis gave nbCI. ClIo Rubidium ____ . ___ . ___ ;3\)'85 30'15 Chlorine ._. _ .. ______24'08 25'04 Iocline ____ . ______44'08 4+-79 RbB}'. Bl',. This can be prepared by adding 4tJ g. of bl'omine to 45 C.c. of an aqueous solutioll containing 50 g. of rubidium bromide, warming gently until bromine dissolves, then cooling. It usually forms a mass of large, brilliant, red crystals in parallel position. 484 Well8, Wheeler and Perljield-RI.tbidiu1ii and

Calculated for Analysis ga ve RbBr, Br., RllbidinnL ____ . _. ____ 25'86 26'26 Rl'Ollline ______. _____ 73'09 737a

RbCl, Bro' This body is prepared by adding brollline to a warllJ satn rated solution of rubidium chloride until some bromi~e re­ mains undissolved and cooling to a low temperatnre, The compolind crystallizes well, but it is the most unstable of the seven rubidium trihalides that have been pl'epared and although it was not fully dried, the sample used for ;nalysi~ suffered a considerable aUlfllmt of decomposition, Calculated fo!' Aualysis gave RbCl, Br., Rubidium . ______32'57 30'42 ChlOl,inc ______14'46 14'44 12'63 Bromine __ . ______49'04 49'40 56'93 In one attempt to prepare this trihalide too much water was used and it was necessary to evaporate off the bromine and concentrate the solution, This operation was repeated several times, after fresh additions of bromine, hefore the propel' con­ ditions were arrived at, and the product finally ohtained was contaminated with RbBr, HI'" as is shown by the following analyses: Calclllated for Calculated for l<'ouud, RbOl, Dr., RbBr, Br., Rubidium _____ 28'78 30'42 20'26 Chlorine ______7'60 6'94 J2'O::! 0' Bromine . __ _ __ 110'92 (j 1'37 56'93 73'73 'Ve have found by experiment that l'l1bidiulll chloride is partly changed to bromide by evaporating an aqueous solution of it with bromine, * This explains the formatioll of the RbBr, Br., Rb Cl, CUll', This compound can he prepared hy adding 33 g, of lkomine to a saturated solution of 50 g, of rubidium chloride, passing chlorine to saturation into the slightly warmed solution, and cooling to a low tempel'ature, The substance is usually de­ posited in the form of very large, light yellow prisms, Calculated for Analysis g-ave RbCl, ClBr, Rubidium ______85'42 35'41 30'15 Chlorine ______:l9'27 28'96 :W'02 Bromine __ . ______31'56 31'89 33'82

* This is in accordance with the results of Potilziu referred to by ~Jendeleje1f in his" Grundlagen del' Chemie" (German ed., 1891) p, 538, Potassi1t1n Trihalides, with their C1'ystallogJ'aphy- 485

Xl_I.,- This body can be made in a few hours by dissolving the theoretical amount of iodine in a hot saturated aqueous solu­ tion of potassium iodide and exposing the resulting solution t~ a winter temperature_ It can also be made as J ollllBon states"" by evaporatino- the solution in a desiccator for a long time_ J ohnson state~ that he always obtained a crop of potassium iodide before the tri-iodide separated_ Vve have never obtained such a product, undoubtedly because we have itlvariably used a sufficient amount of iodine_ It was not considered necessary to make a new analysis of this body_ KEr_BrI_ This compound can be prepared by making a very concen­ boated, warm solution of the calculated amounts of potassium bromide, bromine and iodine, and exposing it for some time to a low t~Hnperature_ The product used for analysis was well crystallized, but it suffered rapid decomposition on exposure to the air_ Calculated for Analysis gave KBr _ BrIo Potassium ______12-21 11-99 Bromine ______51-25 51-61 49-06 Iodine ______30-42 20-11 38-94 KCl_ ClL To prepare this substance, chlorine is passed into a warm mixture of calculated quantities of potassium chloride and iodine in the presence of an amount of watel- insufficient to dissolve the potassium chloride even when hot_ The stream of chlorine is stopped as soon as the iodine has been converted into the monochloride, for otherwise Filhol's well-known com­ pound KCl _CI,I will be formed_ Everything is then dis­ solved by warming and cautiously adding water if necessary and the solution is exposed to a low winter-temperature_ Th-e crystals are very unstable, but apparently not quite as much so as KBr _ErI. Calculated for Analysis gave KCI_ ClI. Potassium ______1.5-29 15-35 16-49 Chlol-ine ______27-53 27-50 29-94 Iodine ______50-37 50-12 53-56 Other Double-halides_ The double-salt CsI _Agi was described in connection with the caesium trihalides as being isomorphous with them as far * Loc_ cit_ 486 Well8, WheeZe?' and Penfield-R~tbidimn, etc.

as the crystals could be measnred. Much work has since been done, without avail, in the hope of obtaining better crystals of this compound. Unsuccessful efforts have been made to obtain me~snrable crystals of all the corresponding silver double halIdes (except the fluorides) with cresiulIl rubidium and potassium. Two or three of these compounds had already been described and it is probable that we could have proven the existence of all the rest of them, but the poorly crystal­ lized products obtained had no interest in this connection and were not analyzed. Repeated efforts also failed to produce from potassium iodide and cuprous iodide a double salt that could be measured.

Theoretical. Arguments were given in the article on the caesium series which have led us to regard the trihalides as belonging to the class of bodies called double halides. We have indicated this view in the present article by using the usual formulre for such compounds. The well-known idea of a linking group of two halogen atoms as an explanation of the structure of double halides was advocated for the caesium trihalides, and, since the rubidium and potassium compounds are entirely analogous, it is unneces­ sary to give their structural formulffi here. We believe, how­ ever, that the trihalides throw some light upon the constitution of the diatomic linking group. Remsen says,* " I cannot see that at present we have any evidence which justifies us in the use of the expression -01=01- rather than -01-01-." If, as we believe, the structure of rubidium h·i-iodide is ex­ pressed by the formula Rb-(II)-1, the strncture of the link­ ing group probably cannot be -17"1-; for in that case a single bivalent iodine atom could do the linking as well as a group of tWI), and we should expect the existence of di-iodides, no evidence of which, or of any other dihalide has been found in the course of an elaborate investigation of the alkali-metal polyhalides. Moreover, with the assumption of bivalent halo­ gen atoms, there would be no difficulty in supposing four halogens to be linked together and the existence of tetra­ halides would be anticipated. Our investigations, however, have shown the existence of only tri- and. pentahalides.t The double linking seems therefore the 1110re probable of the two forms mentioned by Remsen, but it may be added that any union weaker or stronger than the others in the 111olecule, and * Am. Chern. Jour., xi, 312. t The pentahulides will be described in a future article. 487 different jJ'01n them, would also explain the non-existence of dihalides and tetrahalides. Assuming that there is a linking group of two halogen atoms in the trihalides, the view advanced, from a considera­ tion of the cresiulD compounds, that the most stable bodies have idelltical atoms in this group is confirmed by the study of the rubidium and potassium analogues. For, on this assulllp­ tion, all the potassium compounds which could be made con­ tain a gl'OUp of identical atoms, while in the missing rubidium compound they are dissimilar. Sheffield Scientific SchooL lIfarch, 1892.