78 HAGA : PEROXYLAMINESULPHONATES AND X.- Perox ylami.rzesul’homt es cum! Hydmx y7amirLetri- sdphonutes (Sulphazilntes and ~~etasul~fzaxl,loctcs). By TAMEMASA HAOA, D.Sc. PERHAPS the most interesting of the sulphazotised salts discovered by Fremy (Ann. China. Phys., 1845, [iii], 15, 408)are the two which result from the oxidation of one or other of the potassium hydroximino- sulphates (hydroxylaminedisulphonates) in aqueous solution by either silver oxide or lead peroxide or some other reagent. One of these products is the very unstable salt which he named sulphuxilate. This substance is remarkable for crystallising from its aqueous solution, which is of an intense bluish-violet colour, in brilliant golden-yellow needles, very slightly soluble in ice-cold water, but dissolving easily in hot water.It can seldom be preserved for any length of time, and gives a disagreeable odour to the skin, like that caused by manganates and ferrates. According to Fremy, it is easily fusible, but that is a mistake ; it is the products of its decomposition which melt. The other salt, his metasuZphaxiZate, is also sparingly soIuble in cold water, but is colourless and has considerable stability. It crystallises inHYDROXYLAMINETRISULPHONATES. 79 rhombic prisms which are so well defined that Fremy describes the compound as being the most beautiful of all the sulphazotised salts. These crystals appear to be isomorphous with those of potassium 5/6-normal hydroximinosulphate and the two salts can hardly be separ- ated by crystallisation. Its solution is quite neutral and gives a precipitate with basic lend acetate only.Premy expressed the com- position of the sulphazilate by the formula H07NS2K2 (here written with the atomic proportions now in use), which is ouly incorrect in iucluding hydrogen (Claus). He gave the composition of the meta- sulphazilate correctly as H,U,,N,S,K,. Claus (AnnaZen, 1871,158, 205), who gave details for the preparation of the coloured salt by lead peroxide, proved that in its production from potassium 5/6-normal hydroximinosulphate it is not necessarily accom- panied by sulphate ; he also demonstrated with tolerable certainty that it is the sole product of the oxidation; and found that it passed spon- taneously into the colourless salt, together with a fixed quantity of sulphate and a gas which is apparzntly nitrous oxide (compare page 89 of this paper).H e recognised the sulphonate constitution of both salts and gave to the coloured salt the name and formula ‘ oxysulph- azotate,’ (S03K),NL\N(S0,K), and to the white salt the name and formula, trisulphoxyazoate,’ O:N(SO,K),,H,O, that is, with the nitrogen quinquevalent in both formulae. It will be seen that, even empirically, these formulz differ a little from Premy’s. Claus also sought for, and, as he believed, obtained, the primordial sulphazotised salt which he formulated at first as SO,,NO,K (op. cit., 213), and afterwards as O:N(SO,K) (Ber., 1871, 4, 508). This would be the analogue of the trisulphoxyazoate,’ with the nitrogen only trivalent; but the existence of this salt has since been disproved (Trans., 1900, Raschig (Annulen, 1887, 241, 223) Laving found that the white salt, in a boiling and feebly acid solution, may yield sulphate to the extent of two-thirds of its sulphur, along with, apparently, hydroxyitminosulphuric (hydroxylaminemonosulphonic) acid, has recognised that i t behaves as a derivative of hydroxylamine.But because it does not decompose when in strongly alkaline solution, he will not allow that it is that base trisulphonated. He modifies Claus’.y two formulae, writing that 0 OH 77, 437). for the coloured salt as (SO&) N-----N(SO,K),, /O\ and that for the \o/ white salt as (S0,K),”O>N(S03K),,2H,0. 0 Hantzsch and Semple have found (Ber., 1895, 28,2744) that, when crystals of potassium 2/3-normal hydroximinosulphate form in a bluish-80 HAGA : PEROXYLAMINESULPHONATES AND violet solution of potassium sulphazilate, they may contain 1-4 per cent.of this salt, apparently in solid solution, and consequently show a bluish-violet colour. These chemists have therefore advanced the view that Fremy’s coloured salt, which they have renamed ‘ nitroxydisulphonate,’ is sulphonated nitric peroxide, the yellow cryst,als of which have double the molecuIar magnitude of the dissolved bluish-violet form, in analogy with the two forms of nitric peroxide itself. The formula for the bluish-violet modification is given in a foot-note as O*N:(SO,K),, in which, therefore, the nitrogen is repre- sented as being trivalent and the oxygen as univalent. A structural formula for the yellow modification is not given, but Raschig’s is rejected, as having two quinquevalent nitrogen atoms in union with aach other, a mode of combination which is without parallel.Raschig’s formula for the white salt is also rejected, but as the simpler one proposed by Claus is adopted in its place, the quinqne- valency of the nitrogen is maintained. On the authority of Schatz- man, and as a result of their own experiments in the case of hydriodic acid, these chemists state that, in acting as an oxidising agent, the coloured salt reverts to hydroximinosulphate. I n 1896, Sabatier (Comyt. yend., 122, 1417, 1479, and 1537; 123, 255) published the results of an investigation of the violet solutions produced by the action of reducing agents on a sulphuric acid solution of nitrososulphuric acid, and suggested that the colour is due to the formation of the acid of Fremy’s potassium sulphazilate.His suggestion is discussed on page 93 of this paper. I n the detailed examination of Fremy’s sulphazotised salts made by Dr. Divers and the author, the results of which have been described from time to time in the Transactions, the sulphazilate and metasul yhaziiate were purposely reserved for separate treatment, because they are distinguished from the other salts in being products of oxidation. In the present paper, the author endeavours to prove (1) that the sulphazilate is an oxime-peroxide (Scholl), or a peroxinze, (SO,K),NO* ON(SO,K),, the first and only inorganic peroxime yet known; (2) that the meta- sulphazilate is a triacykated hydroxylamine, (SO,K),NO(SO,B), being the only compound of this type having an established normal con- stitution (all others, such as tribenzhydroxylamine, being apparently of more complex constitution) ; and (3) that, consequently, the nitrogen in both these sulphazotised salts is only tri’valent, instead of being quinquevalent.From among the several constitutional names which suggest them’- selves for Fremy ’ s provisional ‘ sulphazilate ’ and ‘ metasulphazilate,’ that of perox?jZaminesuZphonccte for the former, and of hydroxylamine- trisulphonate for the latter, have been adopted as preferable. InHYDROXYLAMINETRlSULPHONATES. 81 consequence, i t has been found advantageous in this connection to call the parent salt hydrox~ylaminedisul~onate, instead of the alternative hydyoxirninosulphate, the name usually employed by Divers and the author.It has also been found convenient to treat of the hydroxyl- aminetrisulphonates before the peroxylaminesulphonates, from which they are apparently always derived, Hydyoxykuminelribu Zphonates (Metasui’phaxilates ; 5?’risuZ@hoxyxmntes). Potassium hydroxylaminetrisulphonate is most readily prepared by Fremy’s method, i n which there is no intermediate separation of its parent salt, the peroxylaminesulphonate. Somewhat a1 kaline potass- ium hydroxylaminedisulphonate is gently boiled and shaken with silver oxide or lead peroxide, until the solution, which a t first becomes intensely bluish-violet, just loses its colour. Then, by evaporation and cooling, the filtered solution can be made to yield nearly all its hydroxylaminetrisulphonate.Theoretically, all the hydroxylamine- disulphonate should be converted into hydroxylaminetrisulphonate and nitrite, 2Pb0, + 3HON(SO,K), + KOH = BPb(OH), + 20N(S03K), + KNO, ; but sulphate and nitrous oxide are always produced, usually accom- panied by very small quantities of nitrogen and aminemonosulphonate (aminosulphate). Nevertheless, 86 and 87.8 per cent. yields of the indicated quantity of the salt have been obtained, together with 78 and 85 per cent. of the full amount of nitrite as indicated by the urea method (p. 94). The production of such large quantities of the hydroxylaminetrisulphonate shows the inaccuracy of Claus’s descrip- tion of the changes concerned. According t o that account, which is endorsed by Raschig, no nitrite is formed, and the utmost yield of hydroxylaminetrisulphonate w-ould be equivalent to only 75 per cent.of tbe sulphur of the hydroxylaminedisulphonate. CorLslitution.-Strictly speaking, the product of the triacylation of hydroxylamine with sulphonate radicles can only be a disulphonate, the third sulphonate radicle becoming sulp6atic by its union with oxygen. But the name of hydroxylaminetrisulphonate is sufficiently appropriate for such a compound, since, although a sulphatic salt, it is not actually a sulphate, but a mixed anhydride of acid salts, one being the 2/3-normal hydroxylaminedisulphonate and the other the acid sulphate : (80,K)OH + HON(SO,K), = H,O + (SO,K)*O*N(SO,K),. Lossen (Bey., 1892, 25, 440) has already pointed out that dibenzhydroxamic acid may be regarded as the mixed anhydride of benzhydroxamic acid and benzoic acid, and similarly in the case of other diacylhydroxy1- VOL.LXXXV. G82 RAGA : PEROXYLAMINESULPHONATES AND amines." Nitrososulphuric acid (aitrosyl hydrogen snlphate), the mixed anhydride of nitrous and sulphuric acids, is an example of a mixed inorganic anhydride. But the present salt, as the anhydride of two different acid salts, finds its close analogue in potassium hyponitrososulphate, (SO,K)O(N,OK) (Pelouze's '' nitrosulphate," Trans., 1895, 87, 109s ; 1896, 69, 1610), which is the anhydride of an acid hyponitrite and a n acid sulphate. The two mixed anhydrides agree in being stable in alkaline solution and unstable in acid solution, and in not giving barium sulphate with barium hydroxide or chloride.The evidence that the metasulphazilates have the constitution of hydroxylaminetrisulphonates is simple and direct, and similar to that as t o the constitution of the hyponitrososulphates. In the first place, sodium amalgam decomposes them, apparently quantitatively (.p. 96), into sulphate and normal aminedisulphonate (iminosulphate) : KO*SO,*O*N(SO,K), + 2Na = KO*S02*ONa + NaN(SO,K),, no sulphite being formed. Instead of sodium amalgam, the zinc- copper couple may be used to reduce hydroxylaminetrisulphonates in boiling solutions (p. 97), but in this case the aminedisulphonnte is apt t o hydrolyse during the heating. The result of this reduction of the salts not only proves their sulphatic constitution but bhows also that neither the formula ON(SO,K), (Claus, Hitntzsch) nor this formula doubled (Raschig) can possibly be right, because its acceptance would require that the sodium should act as a "carrier " of oxygen t o the sulphonate radicle.Dunstan and Goulding (Trans., 1899, 75, 792) have found that trialkyloxamines, such as (CH,),N:O, are reduced to frialkylarnines by zinc and acid. Were metaeulphazilates also oxaminic in constitution, they too should be reduced to aminetrisulphonates (nitrilosulphates). Sulphi tee, and even sulphur dioxide, have no action on the hydroxylaminetrisulphonates (p. 98). I n the second place, the metasulphazilates behave as sulphonated hpdroxylamine. They reduce acidified permanganate ; they give up one-third of their nitrogen in the form of ammonia when they are heated with soda-lime (Claus) ; and they can be hydrolysed ultimately into hydroxylnmine and acid sulphate.Although very stable salts in other respects, they cannot, indeed, remain in solution very long or be * Twenty-six years ago, Koenigs (Ber., 1878, 11, 615 and 1588) found that benzenesulphinic and nitrous acids react to form hydroxylaminedibenzsulphinic (dibenzsulphydroxamic) acid, and that this with more nitrous acid becomes a tri- benzsulphinic compound. Preliminary experiments made for the euthor seem t o show that the latter will almost certainly prove to be hydroxylaminetribenz- sulphizlic acid. Its production may probably be expressed by the following equation : 6C,H,*SO,H + 4HO'NO = 2(G6H,'SO,),N'O*(SO,'C,H,) + N,O + 5H10.HYDHOXYLAMINETRISULPHONATES. 83 kept for many months in the solid state without beginning t o hydrolyse. But if small amount of potassium or sodium hydroxide or, much more conveniently, of ammonia is added to their solution, they are permanent even for years in closed vessels.The other less sulphonated hydroxylamines have no such stability, but always revert more or less to sulphite and either nitrit.e or nitrous oxide. The hydrolysis is expressed by the equation : (SO,K)ON(SO,K), + 3H20 = 3H2S0, + HOaNH,. Taking into corrsideration their water of cry stnllisation, the potass- ium and the ammonium hydroxylaminetrisulphonates can only be written with doubled formulae, thus in some degree supporting Raschig's action in doubling Claus's formula for the former salt. But a cryoscopic measurement (p. 100) of the molecular magnitude of the sodium salt has shown that the simple formula is correct.hydl'oxYIamil2etrisu~honute, 1 2 (SO,K)ON( S0,K) 2, 3H,O, hitberto the only known salt, occurs in flattened, monosymmetric prisms, measurements of which have been made by Fock (Raschig) Its solubility in water a t 1 8 O is one in 25.37 parts. It is neutral to phenolphthalein, litmus, met hyl-orange, and other indicators. When slowly heated t o 100-120' in the air, i t loses some of its water of crystallisation, and is then hydrolysed by the remainder, acting together with the moisture of the atmosphere, so that at first it loses in weight and then gains. The residual mass is strongly acid, owing to the presence of acid sulphate. It has not been found possible to avoid hydrolysis and to obtain the anhydrous salt, even when the compound is very gradually heated in a current of dried air, after having already been exposed in a desiccator at the ordinary tempera- ture.Its water, therefore, could only be determined by difference. As expressed by $he foregoing formula, which agrees with Fremy's empirical formula, i t is certainly 3/2H,O, although Cllaus made it out to be 1H,O only. In his paper, five concordant analyses of the anhydrous salt, besides four analyses of the hydrated salt, are given ; and so far from reference being made to any difficulty being experi- enced in rendering the salt anhydrow, i t is stated that the water of crystallisation easily escapes at 100'. But i t is important to note that his four determinations of the water give numbers which are all somewhat higher than those required by his calculation, although the salt occurs in large, clear, non-deliquescent crystals, and that the figures thus calcu- lated are but little if any higher than those obtained by the author in two direct determinations of the loss of water by heating, in which the residues were always acid and therefore contained water.The salt has also been analysed by Raschig, but his results are not decisive Potassium ( P o 99)- a 284 HAGA : PEROXYLAMINESUJ,PHONATES AND Sodium H y ~ ~ o x y l a m i n e t r i s u ~ ~ o n a l e , (S03Na)ON(S0,Na)2, 2H20.- This salt, now prepared for the first time, is obtained by boiling zt solution of 2/3 normal sodium hydroxylaminedisulphonate and i t s equivalent half-molecule of sodium hydroxide with lead peroxide.It is more difficult to purify from accompanying salts than t'he potassium salt, but by very cautious addition of sulphuric acid, these salts may be con- verted into sulphate, which can be easily separated from the hydroxyl- aminetrisulphonate by freezing. It crystalliscs in aggregates of small, tabular, monoclinic crybtals (p. 99). The solubility of the salt is considerable, one part requiring only 2.83-2.85 parts of water a t 21.5'. Like the potassium salt, i t is neutral to indicators, and, when heated, hydrolyses in its water of crystallisation. Ammonium HydroxyZain.inet&ulphonate, H3,O23N$, or 2( S03NH,)ON(S03MH,),,3H20. -The analysis of this salt confirms the view that the amount of water present in the potassium salt is greater than that found by Claus.The ammonium salt forms thick, rhombic plates and prisms, similar to those of the potassium salt and probably isomorphous with them. But goniometric examination was impracticable, for although some faces were 7-9 mm. long, others were too imperfectly developed for determination. The salt is neutral to litmus and methyl-orange, and generally like the potassium salt, but it is exceedingly soluble in water, one part dissolving in 0.61 part a t 16'. The salt examined was prepared by digesting the basic lead salt with ammonium carbonate and evaporating the solution on the water-bath until it had almost lost its alkalinity, and then concentrating it further under reduced pressure over solid potassium hydroxide.I t s nitrogen and sulphur were found to be in much closer agreement with the formula showing 3/2 molecules of water; but here again the difference between the numbers for the two formulze is not very great (p. 100). Hydroxy- lead Hydroxy Za~~irLetrisulpl~onate , ON(S0,PbO .PbOH),,3H2O. -This tetrabasic and very insoluble lead salt, which appears to be the only insoluble hydroxylaminetrisulphonate, was prepared by pouring a warm solution of the potassium salt into excess of carefully prepared basic lead acetate solution. It is a chalky powder readily decomposed by a solution of an alkali carbonate (p. 100). PeroxyZamines Lclphottates (SuZphuxiZates ; Oxysulphazotates ; Xitroxydi- Only silver oxide and lead peroxide have, as yet, been used in the preparation of a peroxylaminesulphonate, but many other oxidising agents produce the violet coloration, thus indicating the conversion of su Zphonates) .HY DROXY LAMIN ETRISULPHONATES.85 a hydroxylaminedisulphonate into a peroxy laminesulphonate, as was pointed out by Fremy ; even chlorine when used in limited quantityis able t o produce this change. Ozone is an excellent reagent, rapidly producing a strong solution of the peroxylaminesulphonate when it is passed into a faintly alkaline solution of the hydroxylaminedisulphonate. Nitrous fumes are absorbed by an ice-cold solution of this salt, which assumes a dark-brown colour, and this solution, when rendered alkaline, slowly acquires the violet colour of the peroxylaminesulphonate. Also, when an ice-cold solution of potassium hydroxylaminedisulphonate and nitrite is barely acidified (preferably with sulphur dioxide), similar effects are produced. The temporary production of a violet colour is frequently observed in experiments made with the compounds of potassium nitrite and potassium hydroxylaminedisulphonates (Trans., 1900,77, 432).Hydrogen peroxide, potassium ferricyanide, potassium permanganate, and alkaline cupric solutions do not interact with a hydroxylaminedisulphonate. Even freshly precipitated mercuric oxide has no action on it, although the oxide is more quickly affected by light when suspended in a solution of the salt. In preparing potassium peroxylaminesulphonate, Fremy showed a preference for the use of silver oxide, whilst Claus, who assumed that the action of silver oxide was apt to proceed too far, preferred lead peroxide.Silver oxide gives a somewhat better yield and none of the silver goes into solution, whereas a little of the reduced lead peroxide dissolves and renders the salt impure. But the dissolved lead is readily removed, and the lend peroxide presents the advantage of being a t hand when wanted, whilst the silver oxide has to be prepared each time and the metal afterwards recovered. Lead peroxide has therefore been used in the present research. Fremy used either the 2/3- o r the 5/6-normal potassium hydroxyl- aminedisulphonat e as the source of the peroxylaminesulphonate ; Claus used only the latter, and Raschig chose the former. The advantage lies with the 5/6-normal salt, for, when prepared from a less alkaline salt, the peroxylaminesulphonate proves to be less easily purified and consequently less stable.The 5/6-normal salt is always so far hydrolysed in dissolving that it is converted into the 2/3-normaI salt, the potassium hydroxide beiug left in solution, as noticed by Claus. The presence of free alkali, however, moderates the action of the oxidising agent, and to such a n extent that a sufficiently concen- trated solution of the very soluble normal sodium hydroxylamine- disulphonate is not attacked a t all by lead peroxide. Apparently, therefore, lead peroxide acts as an acid oxidiser, in the form ot plumbic anhydride, as huggested by Fremy. The salt, which must be prepared just when it is wanted, is produced by mixing about 6 grams of the 5/6-norma1 hydroxylaminedisulphon-86 HAQA : PEROXYLAMINESULPHONATES AND ate (or the same amount of the 2/3-normal salt together with a small quantity of potassium hydroxide) and a little more than the same weight of lead peroxide (or of the silver oxide precipitated from a little less than the same weight of silver nitrate) and making up with water to 25 C.C.The mixture is agitated for 15 minutes in water near to, but not above, 40'. Then the solution is decanted without delay, treated with carbon dioxide (when lead peroxide has been used), and filtered, before cry stallisation sets in. The solution should therefore be kept warm up to this point. When it has remained some hours in an ice-box, almoht the whole of the peroxylaminesulphonate will have separated as a crust of minute, yellow needles.These can be recrystal- lised, but not without material loss, from hot water made slightly alkaline with potassium hydroxide. When, as appears to have been the case with Fremy, much hydroxylarninedisulphonate has been left unoxidised, some of this will be found with the peroxylaminesulphon- ate, from which it can hardly be wholly separated by recrystallisation, its crystals remaining coloured by the peroxylaminesulphonate, as observed by Hantzsch and Semple. Any close determination of the yield cannot be made directly, since the salt can rarely even be roughly weighed before decomposition sets in. Its amount has therefore to be estimated by letting i t decom- pose, igniting the residue with ammonium carbonate, and weighing the potassium sulphate. In this way, the yield of separated salt was found t o be a very little over three-fourths of the calculated quantity when silver oxide was used; and a little less than two-thirds when lead peroxide was taken.But by indirect means the amount of the salt actually produced can be shown to be much higher than this. As already mentioned (p. Sl), the exhaustive oxidation by lead peroxide of a hot solution of hydroxylaminedisulphonate has given nearly 88 per cent. of the calculated quantity of hydroxylaminetrisul phonate, a fact which signifies that a t least as much peroxylaminesdphonate as is equivalent to this percentageof the total sulphur must have been formed, since its production is intermediate to that of the hydroxylaminetri- sulphonate.Potassium peroxylaminesulphonate is very unstable in water and very slightly soluble in the cold. I n N/iO-solution of potassium hydroxide, which fairly represents its usual mother liquor, it is more stable, but still not very soluble; 100 parts a t 3O dissolve only 0.62 part of the salt, and a t 29Oonly 6.6 parts (p. 101). It interacts in solution with normal potassium sulphite and then produces hydroxyl- aminetrisulphonate and hydroxylaminedisulphonate, evidently in mole- cular proportions (p. l O l ) , this change being a fact of great theoretical importance. Its chemical activity is manifested in oxidising certain easily oxidisable substances and being thereby reduced to its parentHYDHOXY LAMINETltISULPHONATES. 87 salt, hydroxylaminedisulphonate.Although i t liberates iodine from hydriodic acid, i t fails t o oxidise hydrochloric acid. When tho latter acid in concentrated solution is poured on the solid salt, it sets up the same decomposition as that which occurs spontaneously (p. 88). But here, as the rise of temperature is moderated, definite although minute quantities of aminomonosulphonate and of hydroxylamine (not i t s sulphonate) can be found. The salt has practically no action on alcohol ; nitrous and sulphurous acids rapidly reduce it, so also does sodium amalgam, first t o hydroxylaminedisulphonate (as already observed by Schatzman), and then this salt passes slowly but corn- pletely into aminedisulphonate (iminoaulphate). Clean granulated zinc slowly reduces the salt, but copper does not.The spontaneous decomposition of the salt may, however, easily be mistaken for its slow reduccion by a reducing agent, since in tbis case also, as will be pre- sently described (p. 89), hydroxylaminedisulphonate is produced. The difference is readily detected by testing for nitrite, which is produced only in the spontaneous decomposition of the salt. Manganese dioxide very slowly decomposes it, causing a minute effervescence ; lead peroxide is inact,ive. Potassium permanganate is reduced t o green manganate. Clean filter-paper, unlike the paper in use i n Fremy's time, does not affect it. Part of the instability of peroxylaminesulphonates must be attri- buted to the presence of oxidisable impurities. Thus, Fremy noticed the decomposing action of atmospheric dust ; whilst nitrite, another impurity liable t o be found in the salt, also greatly increases its instability.Acids hasten the decomposition of the salt ; alkalis retard it. When drained on the tile from an alkaline solution, the salt may, under cover, remain undecomposed for two hours or more; but if washed on the tile and thus deprived of the traces of its adherent alkaline mother liquor, it will decompose i n a very few minutes. Nevertheless, on one occasion some of the salt thus purified mas kept on a tile for 11 months in a desiccator, and only then decomposed through a n accident. This, however, must be regarded as a very un- common experience. The sensitiveness of potassium peroxylamine- sulphonate t o acids has been recorded by others, but has been some- what exaggerated.When the salt is free from every trace of nitrite, its cold acidified solution may remain coloured for 40 minutes. An alkaline solution may not lose all its colour when kept in a closed vessel for more than a month. If sufficiently pure, the solid salt ruaj be preserved for a day or so under water rendered slightly alkaline with potassium hydroxide. The nature of the decomposition of potassium peroxylaminesulphon- ate occurring in the absence of alkali has already been examined, although in all cases very imperfectly, by Fremy and by Claus, and inss HAGA : PEKOXYLAMINESULPHONATES AND the presence of alkali by Kaschig. According to Fremy, the solid salt decomposes explosively when heated; when exposed to the air, it becomes strongly acid ; and when heated in solution, it yields sulphate and a gas mistaken by him for oxygen, but which was really nitrous oxide.H e was also mistaken in stating t h a t it melts readily and that when left in a closed bottle it evolves nitric oxide. Claus found that, whether in the solid state or in solution, whether when cold or moderately heated, the decomposing salt yields hydroxylaminetrisul- phonate and nitrous oxide, together with acid sulphate equivalent to one-fourth of its sulphur, according to the equation 2K,N,S401, + H,O = 2KHS0, + 2K,NS,O,,, + N,O. Rwchig has confirmed Claus's statements and also states that the solid salt or its solution also decomposes in this way even when left in contact with alkali. H e also adds that its solution when acidified is decolorised in a few minutes, whilst in the presence of alkali it can in some cases be heated to boiling without change.All these statements by Claus and Raschig require t o be modified in order that they may accurately describe the behaviour of the salt, and even then they fail to indicate the primary change which the decomposing salt undergoes. Thus the acid sulphato produced is seldom equal to one-fourth of the total sulphur, although it may be so, as twice found by Claus, and indeed also in the present investigation, but only when the salt had been used with too small a quantity of water to dissolve i t all on warming. 0 wing to this fact, only a part of the nitrogen, which does not become hydroxylaminetrisulphonate, appears as nitrous oxide, along with a small amount of free nitrogen.Solid potassium peroxylaminesulphonate is too unstable when dry and free from alkali t o exist many minutes without rapidly and almost explosively decomposing. I n this decomposition, slight white fumes of ammonium salt (probably pyrosulphite and pyrosulphnte), nitrogen and nitrous oxide, and a small quantity of sulphur dioxide are given off, whilst the residue, when the mass of the salt has been at all considerable, gets very hot (above 300'2) and melts. This residue consists of potassium sulphate (principally pjrosulphate) with a very little ammonium salt. Sometimes a trace of amine- moiiosulphonate can be detected by the mercuric nitrate test ; also a trace of hydroxylamine (or other substance reducing alkaline cupric solution), but none of the other sulphonates, the temperature having been too high to leave these substances undecomposed.The true products of the spontaneous decomposition of a peroxgl- aminesulphonate are only found (in company with small quantities of apparently secondary products) when the salt is heated t o boiling with enough water to dissolve it, and i n presence of sufficient alkali t o prevent both the acidification of the solution during the decompositionHTDROXPLAMINETRISULPHONATES. 89 of the salt and also the secondary changes which would result from acidification. The alkali does not appear t o modify the nature of the primary change, although i t distinctly increases the stability of the salt, as already mentioned (p. 87). When carriedout in the foregoing manner, the decomposition of a peroxylaminesulphonate proceeds largely in such a way that not only do three-fourths of the sulphur of the salt, as suggested by Cllaus and by Raschig, together with one- half of its nitrogen, come out as hydroxylaminetrisulphonate, but the rest of the sulphur and one-fourth of the nitrogen become hydroxyl- aminedisulphonate again, whilst the remainiog one-fourth of the nitrogen appears as nitrite, although some nitroiis oxide and sulphate, besides minute and uncertain quantities of other substances, are always produced (p.92). This result explains the production of the large quantities of acid sulphate and nitrous oxide observed by Claw and Raschig, for the nitrous acid when not neutralised by alkali interacts with the hydroxylaminedisulphonate and yields acid sulphate and nitrous oxide (Trans., 1900, 77, 433).The regeneration of hydroxylaminedisulphonate in the spontaneous decomposition of a peroxylaminesulphonate accounts for the fact, met with in the present investigat,ion, that much more hydroxylamine- trisulphonate is obtainable Ly heating hydroxylarninedisulphonate in solution with excess of lead peroxide than can be derived from the decomposition (out of contact with lead peroxide) of the peroxylamine- sulphonate equivalent to that quantity of hydroxylaminedisulphonate (p. 81). For in the presence of lead peroxide, that hydroxylaminedi- sulphonate which is regenerated by the independent decomposition of the peroxylaminesulphonate is oxidised again to more of this salt, to be again decomposed in the same way, until the whole of the disulphonate has become trisulphonate and nitrite, except that part of it which is lost as sulphate and nitrous oxide.Remembering that 4(SO,K),NOH gives 2( S03K)4N202, the larger yield of hydroxylamine- trisulphonate which should result mill be seen, on comparing the equation on page 81 with that just given, to be theoretically in the ratio 4: 3. It will now, too, be evident, on reference to Claus’s memoir, that his incomplete knowledge of the nature of the decom- position of the peroxylaminesulphonate led him to object too much to Fremy’s account of the action of the oxidising agent in producing hy droxy lamine tri sulphonat e. Conatitution.-The constitution of a peroxylaminesulphonate as a sulphonate was recognised by Claw, and is deduciblz from the fact t h a t it is formed by the dehydrogenation of hydroxylaminedi-90 HAGA : PEROXYLAMINESULPHONATES A N D sulphonate.The problem of its constitution as a nitroxy-compound remains to be solved, and the description just given of the potassium salt amounts to a demonstration, first, that its constitution is that of a peroxide and therefore of a peroximide; and, secondly, that its nitrogen is trivalent. Among the facts bearing on its constitution as a peroxylamine, that ip, as a derivative of H,NO*ONH,, are, first, those of its mode of formation. The 2/3-normal hydroxylaminedisulphonate hses its two hydrogen atoms, at the ordinary temperature and when i t is in aqueous solution, by the action of ozone, lead peroxide, silver oxide, and a variety of other substances : but not, however, by oxygen itself; for Raschig's observation, that n solution of hydroxylaminedisulphonate when exposed to the air may assume a slight violet colour, applies in reality only to the case where the 2/3-normal salt is contaminated with nitrite, the pure salt never oxidising nor colouring in this way. The interaction with lead peroxide points clearly either to the peroxide constitution or, but with much less probability, to a rise in the combining power of the nitrogen to quadrivalency.The fact of the ready reversion, ah the cornwon temperature and in solution, of a peroxylaminesulphonate to a hydroxylaminedi- sulphonate by acting as an oxidising agent is equally strong evidence of the same constitution.This reversion is also quantitative to an extent that admits of its being used to estimate the amount of the salt present in a solution (Schatzmann, Hantzsch and Semple). Its combination with a molecule of normal sulphite (p. S6) affords convincing evidence to the same effect, since it is effeckecl through the oxygen atoms of the peroxylaminesulphonate : (SO,K),NO*ON(SO,K), + K*SO,K = (SO,K),NOT< -t- (SO,K)ON(SO,K),. This interaction will be again discussed on page 91. Inferen- tially in favour of the peroxide constitution are also the odour which the peroxplaminesulphonates impart to the skin, their colour, and their decomposition into nitrous acid and sul phonat'ed hy droxylamines. So soon as it is recognised that peroxylaminesulphonates are per- oxides, all doubt is removed as to the valency of their nitrogen, which then can be only that of a triad.Contrariwise, when such a con- stitution is not admitted, the nitrogen of a peroxylaminesulphonate, with equal certainty, cannot be trivalent. I n order, therefore, to strengthen the conviction that the peroxylaminesulphonates are indeed peroxides and peroximides, it becomes important to state theHYDROXY LAMINKTRISULPHONATES. 01 reasons against admitting the nitrogen of these salts to be quadri- valent or more than trivalent. To begin with, i t is extremely improbable that oxidation by lead peroxide, silver oxide, or ozone should raise the valency of the nitrogen to only quadrivalency and not to quinquevalency, and that it should raise it a t all without converting the sulphonate t o sulphate radicles.Neither Claus nor Raschig assumes that it does, for according to them the nitrogen of the hydroxy lamined isulphonate is itself quinquevalent, But there are also two strong reasons for rejecting the assumption that the valency of the nitrogen is raised by the oxidation of hydr- oxylaminedisulphonates to peroxylaminesulphonates. One of these is the nature of the products of the spontaneous decomposition of a peroxylaminesulphonate. These products, in so far as they contain nitrogen, are all trivalent nitrogen compounds, namely and in the main, hydroxylaminetrisulphonate, hydroxylaminedisulphonate, and nitrite; if nitrous oxide is also recognised, that fact will not affect the argument. No nitrate can be found among these products (p.102). It is, of course, the establishment of the trivalency of the nitrogen of the first-named product which has really settled the matter. B u t as it is only as yet on the chemical work of Divers and the author (Trans., 1894, 65, 523), that the adoption of the trivalency of the nitrogen in hydroxylaminedisulphonates can be based, the result of a determination by a cryoscopic method (p. 100) of the molecular magnitude of the normal sodium hydroxylaminedisulphonate may be adduced in support of it. This result shows that the molecule of the salt contains but one atom of nitrogen (necessarily, therefore, trivalent), and not two atoms as had been represented by Claus and by Haschig. Now, the spontaneous decomposition of a peroxylaminesulphonate can only be hydrolytic, and is therefore one not affecting the valenoy of the nitrogen ; or, should this be contested, i t can still be asserted that at least this decomposition cannot be interpreted as a change involving a diminution in the valency of the nitrogen.The other reason against the belief that the valency of the nitrogen changes when a hydroxylaminedisulphonate is oxidised t o a peroxpl- aminesulphonate is that of the production of the two compounds of trivalent nitrogen, the hydroxylaminedisulphonate and hydroxylamine- trisulphonate, by the union of a peroxylaminesulphonate with a normal sulphite (p. 86). These two reasons for rogarding the nitrogen of a peroxylaminesulphonate as trivalent seem to be conclusive, and there- fore support the view that these salts are constituted as peroxides or peroximides.Since the sodium salt is even more unstable than the potassium salt, the determination of thu molecular weight of a peroxylamine- sulphonate has not been possible. It would seem better to modify92 HAGA : PEROXYLAMINESULPHOKATES AND Hantzsch and Semple's suggestion concerning the molecular weights of the two forms of t h e potassium salt (p. SO), t o the extent of giving the simple formula, (S0,K),N,02, t o the violet form, and reserving the double formula, or even a higher multiple of this, for the yellow form. P~oducts of Decomposition.-Without further experiments than those described on pages 94 and 106, the number of the products and the great variations in their proportions are such that the nature of the spontaneous decomposition of a peroxylaminesulphonate cannot yet be fully determined.But its general character can be indicated, now that the constitutim of both peroxylanlinesulphonates and hydroxyl- aminetrisulphonates has been determined. It can hardly be doubted t h a t the molecule of peroxylaminesul- phonate becomes halved by hydrolysis and converted into the hydroxyl- aminedisulphonate, always found i n abundance, and the hydyoperoxyz- uminesui'phonute, as yet undiscovered because incapable of continued existence, 1 h u s : (S03K)2NO*ON(S0,K)2 + H,O = (SO,K),NO*OH + H*ON(S0,K).3. It is already known (Trans., 1889, 55, 765 ; 1894, 65, 539) tbat, in the presence of a1 kali, the nitroxy-radicles oE a hydroxylaminesul- phonate tend to separate from the sulphonate radicles.Such a tendency, exercised in the presence of undecomposed peroxylamine- mlphonate, will lead to the production of hydroxylarninetrisulphonate and nitrite in the case of hydroperoxylaminesul phonate, and of the former salt and hypouitrite in the case of hydroxylaminedisulphonate, thns : (SO,K),N,O, + (SO,K),NO*OH = Z(SO,K),NO(SO,K) + HO*NO ; (S03K),N202 + (SO,K),NOH = 2(SO,K),NO(SO,K) + h(HON),. When the three equations are combined, the intermediate products disappear and the following equation is left, 6(S0,K),N202+ H,O= G(SO,K),NO+ 2NO,H+N,O ............ (1) or, leaving the comparatively stable hydroxylaminedisulphonate unchaoged, 2(SO,K),N,O, + H,O = 2(SO,K),NO + (SO,K),NOH + NO,H .. (2) It is fairly certain that the sulphate which, in grei-ttlg varying although never very large quantity, is always produced, does not come from the hydrolysis of the salt itself or from t h a t of either the hydroxylaminetrisulphonate or hydroxglaminedisnlphonate derived from it.For the trisulphonate is remarkably stable in the presence of alkali, and the disulphonate, although unstable in its presence,HYDROXYLAMIN ETRISULPHONATES. 93 yields not sulphate but sulphite. As this is also true of hydroxyl- aminemonosulphonate, i t may be assumed to be so in the case of peroxylaminesulphonate. The sulphate should therefore have another origin, which may well be taken t o be the decomposition of the hydroperoxylaminesulphonate in circumstances i n which i t fails to interact with peroxylaminesulphonate, perhaps because the tempera- ture of the solution is too lorn.I n that case, i t will naturally hydro- lyse, one half becoming hydroxylaminedisulphonate by oxidising the other half into sulphate and nitrous acid, B(SO,K),NO*OH + H,O = (2S0,KH + N0,H) + (SO,K),NOH. Or, it may well hydrolyse wholly into sulphate and nitrous oxide, 2(SO,K),NO*OH + H,O = 4S0,KH + N,O. These equations combined with the primary equation give, 2(S0,K),N,02 + 3H,O = 2S04KH + 3(S0,K)2NOH + N0,H . .. (3) 2(SO, K),N2O2 + 3H,O = 4S0,KH + Z(SO,K),NOH + N,O . . . . . . (4) An equation t o account for the production of nitrogen, and another for that of aminemonosulphonate are easily framed : (SO,K),N,O, +2H,0=4S04KH +N, .............................. (5 ) (SO,K),N,O, -I- 3H20 = 3S0,KH + (SO,K)NH, + NO,H... . . . . . . (6) A justification of the lower of these equations and a n illustration of the nature of the change, expressed by i t are to be found in the pro- duction of aminemonosulphonate from hydroxylaminedisulphonate when decomposing in presence of copper sulphate (Trans., 1900, "7, 978). By combining these six equations in different ways, the various results obtained can be explained (p. 107), although the conditions for the occurrence of one mode of decomposition more than another are not yet ascertained. Sabutier's Bluish-violet Acid.-Sabatier has studied the nature of the bluish-violet colour produced in a solution of nitrososulphuric acid (nitrosyl hydrogen sulphate) in the monohydrate of sulphuric acid by sulphur dioxide, and in other ways (p. SO), and has found this colour t o be more closely like that of a solution of potassium sulphazilate than the colour of the latter is like that of a solution of potassium permanganate.Oa this ground and from a consideration of the circumstances which give rise to the colour, he has suggested that i t is due t o the presence in the solution of the acid of Fremy's salt, constituted according to the formula ON(SO,H),. Sabalier may be right, but there is much to be said against this opinion. Firstly, the tints of the two coloured solritions are not so similar asthe asserts.91 HAGA : PEROXYLAM"ESULPH0NBTES AND Secondly, certain striking contrasts may be observed in the chemical character of the two solutions. Potassium peroxylamineoulphonate is produced by the action of lead peroxide and is not attacked by it, whereas the coloured acid solution is a t once oxidised by lead peroxide.Conversely, whilst this acid solution is indifferent towards sulphur dioxide and produced by it, potassium peroxylaminesulphonate is a t once changed by this reagent. Then, again, it has not proved to be possible either t o convert potassium peroxylaminesulphonate into this violet acid solution or to effect the opposite change. Mr. S. Sekiguchi, a Post-graduate of this University, has kindly carried out some experiments in this direction. Making the mixtures very gradually and keeping them cold by ice and salt, he has poured the solution, prepared from nitrososulphuric acid and sulphur dioxide in sulphuric acid, into a solution of potassium hydroxide; and, on the other hand, an aqueous solution of potassium peroxylaminesulphonate into some concentrated sulphuric acid ; in both cases, an almost immediate disappearance of the violet colour results.In the former case, too, the alkaline solution mas evaporated and crystallised, without finding any of the hydroxylaminetrisulphonate which would result from the decomposition of peroxylaminesulphonic acid and might, to some extent, in accordance with its usual stability, escape decomposition. Details of Expeviments. The Exhaustive Action of Lead Peroxide 031 Hgdroxykaminedi- gdphonates.-Potassium 213-normal hydroxylaminedisulphonate was boiled with excess of lead peroxide in about 15 times its weight of water, containing from 1/5 to 2/5 of a molecule of potassium hydr- oxide, until the solution had again become colourless.To the cold filbrate, just enough barium acetate was added t o precipitate all sulphate present ; the filtrate was then evaporated, and the hydroxyl- aminetrisulphonate crystallised out, as far as possible, and weighed. Potassium nitrite, produced in large quantity, was estimated in the mother liquor and alcoholic washings of the crystals of the hydroxyl- aminetrisulphonate by the urea method. Sulphate was found partly in solution and partly in the lead residue, which was extracted alter- nately with dilute nitric acid and potassium hydroxide. The sulphate, both in solution and residue, was estimated, and, in two cases, the soluble lead also, as a measure of the lead peroxide consumed. In one experiment, 73.2 grams of salt gave 58.2 grams of trisulph- onate in crystals, that is, 58.5 mol.of trisulphonate from 100 mol. of dipulphonate, or 87.75 per cent. of the theoretical quantity. Nothing else was determined, and so high a yield of hydroxylaminetrisulph-HYDROXYLAMINETRIYULPHON A'I'ES, 95 onate was only reached by adding alcohol to separate the last portions of the salt from the very concentrated nitrite mother liquor. In another experiment, 125 grams of the disulphonate gave 97.33 grams of trisulphonate, equal t o 57.33 mol. of trisulphonate to 100 mol. of disulphonate, or 86 per cent. of the calculated quantity. The amount of potassium sulphate was 21.5 mol. per 100 mol. of salt takeo, which leaves sulphur for the trisulphonate equivalent to 59.5 mol., as against the 57.33 mol.of crystallised salt. Very much nitrite mas found (37.5 mol. per 100 mol. of disulphonate taken), indicating the production of very little nitrous oxide. The only way t o interpret this large production of nitrite is to assume that, whilst 89-25 per cent. of the salt was oxidised into trisulphonate and nitrite, and only 3 per cent. into sulphate and nitrous oxide, 7.75 per cent. was oxidised into sulphate and nitrite, an assumption which cannot be easily justified. I n an earlier experiment, in which the crystallisation of the tri- sulphonate was only imperfectly carried out, 136.33 grams of the disulphonate gave 84.33 grams of the crystalline product, that is, 100 mol. gave 45.56 mol., or 68.33 per cent. of the theoretical quantity. But since the quantity of sulphate, almost if not actually the only other sulphur compound produced, amounted to only 19 mol.per 100 of disulphonate, the actual yield of trisulphonate can have been little short of 60.33 mol. per 100. The nitrite, as determined by the urea method, was 28.4 mol. per 100 oE disulphonate taken, But the lead peroxide consumed was in this case determined, and made out to be 71.5 mol. per 100 mol. of disulphonate, and this indicates the produc- tion of 31.8 mol. of nitrite. Accepting the mean of these numbers for the nitrite, i t results that about 90 per cent. of the hydroxylamine- disulphonate was converted into trisnlphonate and nitrite, and the rest into sulphate and nitrous oxide, I n an experiment with 35 grams of potassium hydroxylaminedi- sulphonate, in which the crystds of hydroxylaminetrisulphonate were not weighed, 100 mol.yielded 13 mol. of sulphate and 26 mol. of nitrite. Calculating from these quantities, 78 per cent, of the salt was oxidised into trisulphonate and nitrite, 1 5 5 into trisulphonate and nitrous oxide, and 6.5 into sulphate and nitrous oxide. The tri- sulphonate produced will therefore have been about 93.5 per cent. of the calculated quantity, or 62.33 mol. per 100 mol. of disulphonate. Another experiment was made on the normal sodium hydroxyl- aminedisulphonate (Trans., 1894, 65, 546) in dilute solution, 1 P33 grams being taken without any addition of sodium hydroxide, because the alkalinity of the salt was sufficient to protect it. But in this experiment, only the quantities of lead peroxide consumed and of sulphate formed were estimated.Exactly as happened in the experi-96 HAGA : PEROXYLAMINESULPHONA'l'k3 AND ment with the potassium salt, 71.5 mol. of lead peroxide were consumed per 100 mol. of sodium salt. The sulphate amounted to 34 mol. per 100 mol. of salt used, more, that is, than in the experiments with the potassium salt. The calculated quantity of sodium hydroxylamine- trisulphonate was correspondingly lower, 55 mol. per 100 mol. or 82.75 per cent. of the theoretically possible quantity. Reduction, of Potassium ~ydroxylaminetrisul~honnte by Sodium Amalgam.-In the interaction between sodium amalgam and potassium hydroxylaminetrisulphonate in aqueous solution, the two liquids become warm, and the action is soon over if the two are well shaken together. No gas is evolved, and nothing is left in solution but the two salts, sulphate and aminedisulphonate (iminosulphate).The latter is easily recognisable by i t s separating as the very sparingly soluble 2/3-normal potassium salt when the solution is nearly neutralised with a n acid, and also by its nearly insoluble normal mercury-potassium salt (Trans. 1892, 61, 976; 1896, 69, 1629). But the salt was also analysed quantitatively (p. 97) in order to demonstrate i t s nature beyond question. By cautiously adding hydrochloric acid t o the cold solution until i t hns become almost neutral t o methyl-orange, and then pre- cipitating with barium chloride, the sulphate is partially separated from the aminedisulphonate; the latter may then be estimated as sulphate in the filtrate after hydrolysis at 150".I n an experiment carried out in this way on 2.447 grams of potassium hydroxylamine- trisulphonate, the barium sulphate precipitate was washed w i t h cold, and then with hot, water, ignited, and weighed. The sulphate from the hydrolysed aminedisulphonate was treated a s in an ordinary sulphate determination. I n this way determined, 34.79 per cent. of the sulphur came out a s sulphate and 64.88 per cent. as aminedisul- phonate, leaving 0.33 per cent. unaccounted for. I n accordance with t h e equation, the actual numbers should have been 33.33 and 66.67 per cent. respectively. By other experiments, it was, however, estab- lished that some of the aminedisulphonnte WRS precipitated with the sulphate.No doubt, also, some barium chloride was carried down. Potassium hgdroxylaminetrisulphonate, 1.441 grams, was reduced by sodium amalgam, and the solution neutralised and precipitated in the cold by barium chloride as above described. The washed pre- cipitate was then heated for 4 hours a t 150" with dilute hydrochloric acid in a sealed tube. The acid was nearly neutralised, and the barium sulphate collected, washed, and weighed as usual. The fil- trate from this yielded a fresh precipitate with barium chloride, for the barium aminedisulphonate, which was precipitated with the sulphate, had been hydrolysed into barium sulphate and ammonium hydrogen sulphate. Therefore, from the weight of t,he main pre- cipitate of sulphate was deducted that of the small quantity lastHYDIXOXYLAMINETRISULPHONATES.97 obtained, and the remainder taken as sulphate actually produced by the sodium reduction. It amounted to the equivalent of 34.20 per cent. of the total sulphur. The aminedisulphonate in the original filtrate from crude sulphate was determined by hydrolysing and weighing its sulphur as sulphate. To the weight of this was added twice that of the barium sulphate obtained, as just described, from the soluble sulphate extracted by hydrolysing the crude barium sulphate, because twice that quantity represented the total sulphur of the aminedisulphonate precipitated along with the actual sril- phate. This sum was equivalent to 65.75 per cent, of the total sulphur. That these data still deviate from the calculated numbers is no doubt due to the adhesion of a little barium chloride to the sulphate when precipitated in the cold.The barium of this chloride will have rendered insoluble some of the sulphate which should have dissolved out through the hydrolysis of the aminedisulphonate simul- taneously precipitated. There seems, therefore, to be no reason f o r doubting the quantitative accuracy of the equation given on p. 82. An experiment was then tried to see whether closer resdlts could not be got by removing as much as possible of the aminedisulphonate from the solution before precipitating the sulphate, first crystallising out most of it from the nearly neutralised solution, and then removing some of the remainder as the mercury-potassium salt, by digesting the solution with mercuric oxide.This method, however, did not give better results than the preceding. Reduction of Potassium HydroxplccminetrisuZphonccte by the Zinc- coppe?. Couple.-The reduction of the trisulphonate was successfully effected by boiling its solution (to which a few drops of sodium acetate solution had been added in order to protect the salt from hydrolysis) with some zinc-copper couple. But in consequence of the necessity of boiling the solution, hydrolysis of the aminedisulphonate is apt to set in. Some aminedisulphonate prepared by the sodium amalgam method, and another sample, prepared by the zinc-copper couple, were analy sed with the following results : By sodimn. By zinc-copper. Calc. Potassium.. . . . . . . . . . . 30.68 30.72 - 30.89 Sulphur,. , . .. , . , . . . . . 25.09 24.85 25-94 25.30 Nitrogen .. , , . . ... ... 5-32 8-16 6.64 Hydrolysis of cc z~ydroz?/~cci~~inetr~su~~~~o~~cct~.--The complete hydro- lysis of the hydroxylaminetrisulphonates is more difficult to effect than that of any other of Eremy’s salts. I n the quantitative analysis of the salts, it was found necessary to keep the acidified solution for 5 hours a t 180-200°. In the case of the potassium salt, the mean VOL. LXXXV. H98 HAGA : PEROXYLAMINESULPHONATES AND percentage of sulphur then came out as 23.18; at 150° only, for 4 hours, it was 22-55 ; and a t 90-100' for 5 hours, and then 3 hours at 130°, it gave only 22.64. In this case, the nitrogen of the hydroxyl- amine obtained (as measured by the iodine method) amounted only t o 2.5 per cent. I n another case, where the hydrolysis was allowed t o go on for 48 hours at 90-95', and then 2 hours a t 130-134', the nitrogen obtained as hydroxylamine was 2.71 per cent.(79.75 per cent. of total nitrogen). Imact%ty of 8ulphites towards ~ydroxy~c6m~netr~su~~honutes.-~o~ass- ium hydroxylaminetrisulphonate weighing 2 grams, in sufficient water to keep the salt in solution, was left for 3 days with 3 grams of potass- ium metasulphite, rendered slightly alkaline t o lacmoid paper (whilst strongly acid t o litmus). The sulphite was then precipitated by barium hydroxide and the filtrate evaporated. I n this way 1.98 grams of the hydroxylaminetrisulphonate crystallised out. The analysis of the salt thus recovered is given as that of I1 among the analyses of the salt below. Analysis of Potassiwnz ~ y c l i .o x y Z a n t i n e t r ~ s u ~ ~ ~ o ~ t a ~ ~ . - A l ~ h o u ~ ~ thia has been analysed by previous workers, it was necessary to make several careful and full analyses in order t o establish the fact that it contains more water of crystallisation than the proportion stated by Claus and by Raschig. Four separate preparations were analysed. I. 0.4954 substance gave 0.3117 potassium sulpbate ; 0,5088 gave 0.8581 barium sulphate ; 0.3387 gave 0*5680 barium sulphate ; 042881 substance, finely powdered and heated in a current of dried air, first a t 96' and then up to llOo, lost 0.0138 ; 0.2174 lost in this way 0*0101. 11. 0.2272 substance gave 0.1429 potassium sulphate ; 1.0676 treated with sodium amalgam for 34 hours and then hydrolysed at 150' for 3 hours, gave 1.7918 barium sulphate and ammonia = 25-59 C.C. N/10 acid.111. 0.2288 substance gave 0.1451 potassium sulphate ; 0.1003 gave 0.1710 barium sulphate. IV. 0.8505 substance gave 0.5347 potassium sulphate ; 0.34475 gave 0.5799 barium sulphate ; 2.4425, by sodium amalgam treatment and hydrolysis at 150" for 3 hours, gave ammonia = 51.38 C.C. NIlO acid; 0,7650, by the Dumas method, gave 22.3 C.C. moist nitrogen a t l6O and 758 mm.HYDROXYLAMINETRISULPHONATES. Potassium, . . . . . . . . . . . . . . . . . . . . . ) ) I1 .................................. -*d Found, I ) ) 111 ................................... ) ) IV ................................... Mean ........................................ Calculated for 1/1H,O ....................,, 3/2H,O ..................... Fremy .......................................... Claus, found (taking old atomic weights) 2 ) J j $ 7 2 1 Y ) 9 ) $ 9 9 , 9 7 Claus, mean ................................. calc. (taking old atomic weights) Raschig ....................................... ) ) ............................... * ' . - I ), 28'25 28 '24 28'47 28-23 25.30 25.96 28-32 28.02 28.63 28.55 28-67 28'79 25'66 28.88 28-47 28'64 Sulphur. ~ {;;:A: 23-07 23'42 23'12 23'18 23'71 23'19 23'40 Nitrogen. - - 3.37 - { 'g;' 3 -38 3'46 3 '39 3'48 - 1 3'24 23'69 - 23'76 3 *31 23.73 3-28 23.70 3.45 23'64 - 23.38 i - - - Water. (4'89) (4'65) - - - - (4.77) 4 '49 6-52 5-04 5.20 5'01 4-71 4.99 4 -44 - __ - - Claus has also given the results of five closely concordant analyses of the anhydrous salt, and should therefore have experienced no diffi- culty in rendering it anhydrous.I n the attempts to determine total water, recorded above, the residue was always acid in consequence of the unavoidable hydrolysis and fixation of some of the water of crys- tallisation. Amalysis of the Sodium Xalt.-Two distinct preparations of the sodium salt were analysed : I. 0.4910 substance gave 0.2801 sodium sulphate; 0,3845 gave 0,7174 barium sulph&e, after hydrolysis at 200' for 3 hours. Hydro- lysed at only 160' for 5 hours, 0.4548 gave 0.8378 barium sulphate, = sulphur 25.30 per cent. only. 11. 0-6097 substance gave 0.3455 sodium sulphate; 0.1750 gave 0.3268 barium sulphate, after hydrolysis at 210° for 3 hour?; 0.7533 gave 2400 C.C.moist nitrogen a t 760.8 mm. and 17O, = 0.027 nitrogen. A discussion of this matter is given on page 83. S odiu m . Sulphur. Nitrogen. - Found, I .................. 18.50 25.64 Na,S3N0,,,2H,O ......... 18.42 25.62 3.76 ,, I1 ................. 18.36 25.67 3-70 Crystallography of Sodium Hydroxykanzi~etrisul~~~o~ate.--Professor Jinbo has kindly given me the following description of the crystals of this salh, which were examined under his directions by Mr. M. Yatsuki, University Post-graduate. Thick, tabular, monoclinic crystals, about 3 mm. long and 2 mm. wide, elongated in the direction of the vertical axis. The observed faces are of seven kinds, of which b is the largest H 2100 HAGA : PlCROXYLAMINESULPHONATES AND and apparently the plane of symmetry.m, making with b an angle of about 115O, may be taken as a prism ; d, e, and f as pyramids ; c as the base, and g as a positive orthodome. Two other faces in the zone of the orthodiagonal are sometimes observed. A crystal laid flat on the clinopinacoid shows an extinction angle of 30' to the vertical axis, in the acute angle between this and the clinodiagonal. AnuZysis of the Ammonium SaZt.-The total nitrogen ol the am- monium salt was determined by tho Dumas method. Found. 2H,,01,N,Sy,3H,0. H,,O,,N,S,,H,O. Sulphur ......... 27.63 37.36 38.08 Nitrogen ......... 16.06 15.98 16.40 Anulysis of the Basic Lead Salt.-The salt wits dried for analysis in a current of dry air a t 100' in the case of preparation I, and at l l O o in that of 11. The salt was quite free from potassium but contained a trace of acetate.Found. Found. H,Q1,NS,Pb,. Lead ............... 74.41 74.14 74.7 1 ............ 5-79 Sulphur 5.99 - Determination of the Molecular Magnitude of Sodium Hydroxyl- aminetrisu2phonate.-This was carried out by Lowenherds method with melted sodium sulphate crystals (Zeit. physikul. Clam., 1896, 18, 70). Fused sodium Hydrated salt. Anhydrous salt. sulphnte. At. 11. w. 2.479 = 2.2411 59-41 1 - 0.369' 332.20 3.478 = 3.0210 57.520 - 0,510 335.56 OI0NS3Na3 requires 339.37 Molecular 3dagnitude of Normal Sodium HydyoxyZaminediuuL phonate.-HYDROXY L AMIN ETRISULPIIONATES. 101 Fused sodium Crystnllised salt. sulpiiate. At. M. W. 1.220 36.8180 - 0.39' 233 2.023 37.1980 - 0.74 239.6 O?NS,Na, requires 239.3 With this molecular weight the nitrogen is necessarily trivalent.Solubilily of Peroxylaminesulphonate in LVjlO Solution of Potassium Hydroxide.-'l'he purified salt, previously washed on the porous tile with some of the solvent, was shaken with i t for from 15 to 20 minutes, the temperature of the solution being 29'. After Some time, 5 C.C. of the clear solution were withdrawn with a pipette. The rest of the solu- tion, along with the undissolved salt, was left for some hours in ice, when again 5 C.C. mere taken out, the temperature being 3'. The two portions were each weighed and the amount of dissolved salt ascertained by a sulphur determination. It was thus found that 0.163 gram of salt was dissolved in 5.03 grams of its alkaline solution a t 29', and that 0.027 gram was dissolved in 4-980 grams of its solution a t 3'.Interaction of Potcrssiurn Pgrox~lc~nzinesul~~onate am! Nornatd Potassium Su2phite.-To a solution of 3.6 grams of potissium peroxyl- aminesulphonate, containing only a very small quantity of potassium hydroxide, a solution of normal potassium sulphite (neutral to phenolphthalein) was added from a burette, with constant stirring, until the violet colour of the solution was entirely discharged. The change took placs quickly but not instantly. The quantity of sulphite required was only a little more than that indicated by theory. After a short interval, baryta water was added to precipitate the excess of sulphite and the hydroxylaminedisulphonate. The excess of baryt a was removed from the filtered solution by carbon dioxide and the filtered solution evaporated so as to get out as much as possible of the sparingly soluble potassium hydroxylamine trisulphonate.Some more of this salt was precipitated by adding twice the volume of alcohol and leaving the mixture for some time. The total trisulphonate thus separated weighed 2.268 grams, or 81.5 per cent. of the calculated quantity. The barium precipitate was triturated in a mortar with very dilute acetic acid, added very slowly so as to avoid as far as possible having any local excess of acid. When the solution had become neutral to phenolphthalein, the undissolved barium sulphiie was filtered off. Potassium carbonate in slight excess was added and the whole left for a day. Then, the solution, filtered from the barium carbonate and neutralised with acetic acid, was concentrated in a vacuum over sul- phuric acid and mixed with twice its volume of alcohol.In 12 hours, the quantity of precipitated crystalline 2/3-normal bydroxylamine- disulphonate weighed 1 *55 grams, this being equal to 76 per cent. of the102 HAGA : PEROXYLAMINESULPHONATES AND calculated quantity. It mas pure, except for a trace of aminetri- sulphonate (nitrilosulphate), doubtless due to the action of the sul- phurous acid unavoidably liberated in the process of separating the barium sulphite from its own barium salt by acetic acid. It was identified by hydrolysis into sulphate and hydroxylaminemonosul- phonate, and above all by its producing the bluish-violet peroxylamine- sulphonate when warmed with lead peroxide and a small amount of alkali.No nitrite was found in the mother liquor of the hydroxylamine- trisulphonate, showing that the production of the latter salt had not been due to spontaneous decomposition of the peroxylaminesulphonate. Hydroxylaminetrisulphonate and djsulphonate are, in fact, the only substances which could be detected among the products of the inter- action of the peroxylaminesulphonate and sulphite. Since, therefore, the separated quantities of theee products were found to be in approximately molecular proportions, and as these salts are not in- soluble, even in their alcoholic mother liquors, it may be regarded as proved that the interaction which takes place is exclusively that represented by the equation on p. 90. Sporztuneous Decomposition of Potassium Peroxyk~~minesuZ~~~onate.- The principal products of the spontaneous decomposition of a peroxyl- aminesulphonate in hot alkaline solution are easy to recognise.Unless very dilute, the solution yields crystals on cooling and more on evaporation, A t first the sparingly soluble hydroxylaminetrisulphonate alone crystallises, and later on both this and the equally sparingly soluble hydroxylaminedisulphonate. In each case the crystals are characteristic and easily distinguished. The presence of the disul- phonate in the solution is quickly and distinctively indicated, as has just been mentioned, by warming with a small quantity of lead peroxide, which gives it again the bluish-violet colour of peroxylamine- sulphonate. By removing sulphate and hydroxylaminedisulphonate from the solution by barium hydroxide, nearly all the hydroxylamine- trisulphonate can be crystallised out ; the mother liquor containing the nitrite may then be tested in any of the usual ways for this salt$.It was important to know whether any nitrate is formed by the decomposition, and therefore necessary first to get rid of all the nitrite present by a process that does not convert any of it into nitrate. The nitrite was accordingly changed into aminetrisulphonate (riitrilo- sulphate) by adding enough potassium carbonate and then passing in sulphur dioxide’until the solution became acid, a t which point the aminetrisulphonate that had been produced a t once hydrolysed (Trans., 1892, 61, 954). Lastly, by blowing in air until all the re- maining sulphrrr dioxide had been expelled, the acid solution was left free from either nitrite or sulphite, and, therefore, ready for testing for nitrate.The application of the None of tbis salt mas found,HYDROXYI~AMINETRISULPHONATES. 103 process of sulphonating the nitrite to the determination of total nitrogen in solution, is described on p. 104. The testing for aminemonosulphonate (aminosulphate) among the products of decomposition of a peroxylaminesulphonate is not an easy matter. The method adopted was to oxidise all the hydroxylamine- disulphonate by boiling the solution with lead peroxide until i t was again colourless. The nitrite was then oxidised by pouring the solution into potassium permanganate solution to which sulphuric acid had been added. iUercuric nitrate solution then precipitated from it a little oxymercurjc aminemonosulphonate (Trans., 1896, 69, 1649), which, when treated with hydrogen sulphide, left the acid agGin in solution.By evaporation and addition of strong sulphuric acid, the acid was obtained in characteristic crystals (Zoc. cit., 1642), which were sometimes weighed. The quantitative examination of the solution is a troublesome and less satisfactory operation. The peroxylaminesulphonate can hardly be obtained for weighing in the dry and pure state, because of its instability. Therefore, after its composition had been found, from concordant analyses of four different preparations, to be that ascertained by previous workers, the preparation of t h e solution and its analysis after the salt had all decomposed were carried out in the following way. The peroxylaminesulphonate, recrys- tallised two or three times from hot water made alkaline with potassium hydroxide, was drained for a short time on a tile from its mother liquor, and at once dissolved in suitable quantity in water to which had been added a measured quantity of potassium hydroxide, The solution was maintained at the boiling temperature until colourless through the complete decomposition of the salt.The cold solution was then weighed off into four portions: one of 5 per cent. of the whole, for estimating the amount of peroxylaminesulphonate that had been dissolved; another of 15 per cent., for estimating the quantity of hydroxylaminetrisulphonate produced ; a third and a fourth portion, each of 40 per cent., for estimating i n one the quantity of sulphate, and in the other that of nitrite produced.The quantity of peroxyl- aminesulphonate taken was determined by weighing as barium sulphate the total sulphur in the solution. To ensure the hydrolysis of all the sulphonate, the solution was heated with hydrochloric acid in sealed tubes for 4-5 hours at 180-200". To determine the quantity of sulphate which had been produced, very dilute hydrochloric acid was added with constant stirring until t h e solution was only barely alkaline to phenolphthalein, then much ammonium chloride was added before precipitating with barium chloride, in order t o keep the hydroxylaminedisulphonate in solution as f a r as possible. The impure sulphate, washed with ammonium104 HAGA : PEROXYLAMINESULPHONATES AND chloride solution on the filter, was transferred t o a beaker, and digested in the cold with very dilute hydrochloric acid, washed again on the filter with boiling water, and then ignited in the usual way.The hydroxylaminetrisulphonate was estimated by leaving the solution with sodium amalgam for two days, occasionally shaking the two together so as to convert this salt into hydroxylaminedi- sulphonate, and then all the hydroxylaminedisulphonate in the solution into aminedisulphonate (iminosulphate). The mercury having been filtered off and washed, hydrochloric acid was added to the solution until it was only just alkaline t o methyl-orange, and then an excess of ammonium chloride was int'roduced. The sulphate was finally pre- cipitated and treated as before described.Deducting from this quantity of sulphate that which was present before the treatment with sodium amalgam, there remained the sulphate equivalent to one-third of the sulphur of the h-jdroxylaminetrisulphonate, from which the quantity of this salt was calculated. Assuming hydroxylaminedisulphonate t o be the only other sulphur compound produced in the decomposition of the peroxylaminesulphonate-an assumption which is newly exact- the amount of this salt was then calculated as being equivalent to the sulphur not found either as sulphate or hydroxylamine- trisulphonate. The slight error in this assumption is caused by t h e production of very small quantities of aminemonosulphonate (aminosulphate). As to the last-named salt, i t has not been possible t o do more than ascertain that its quantity is usually quite small, although 2 mol.of tho crystalline acid (p. 103) per 100 mol. of hydroxylaminedisulphonate oxidised by the lead peroxide were once actually obtained. I n ot'her words, the amount of sulphur found as aminemonosulphonate was in this instance 1.03 per cent. of that of the hydroxylaminedisulphonate used. To determine the total nitrogen in the solution, the nitrite was completely sulphonated to aminetrisulphonate (nitrilosulphate) by adding enough potassium carbonate for the purpose and then passing in sulphur dioxide until a piece of lacmoid-paper was just reddened (Trans., 1892, 61, 954). Next, the hydroxylaminetri- sulphonate in the solution was reduced by sodium amalgam, as above described, to sulphate and aminedisulphonate.Having thus brought all the nitrogen into aminesulphonates, the hydrolysis of these substances by hydrochloric acid was effected by heating, first, on the water-bath until all t,he sulphur dioxide had been expelled, and then for some hours in a pressure-tube a t 150O. The solution distilled with alkali gave up all its nitrogen as ammonia. The difference between this and that originally present as peroxylamine- sulphonate gives, indirectly, the quantity of nitrogen in the gases, whilst the difference, again, between the total nitrogen in the solutionHYDHOXYLAMINETRISULFHONATES, 105 and the sum of the quantities found as disulphonate and trisulphonate is the nitrogen which was present as nitrite. Although the experimental work in estimating sulphate and hydroxylaminetrisulphonate was performed with great care, no high degree of accuracy in the results could be expected. A test experi- ment was made t o see to what extent the method was imperfect. A solution was prepared by dissolving potassium sulphate, potassium hydroxylaminetrisulphonate, potassium hydroxylaminedisulphonate, and sodium nitrite in water to every 100 C.C.of which 5 C.C. of N/lO solution of potassium hydroxide had been added. The solution was twice analysed for sulphate and trisulphonate in the way described above. The quantities, taken and found, are here given in grams per 100 C.C. Taken. F o iuld. Trisulphonate.. .... 2.580 2.547 2.624 Sulphate ............ 0.347 0.380 0.373 Disulphonate ......0,622 0.629 0551 Nitrite ............ 0.208 0.212 0.217 From this experiment, i t seems that the sulphate may come out nearly 10 per cent. too high, no doubt for two reasons; one that being precipitated in the cold, the barium sulphate retained other salts with i t ; the other and principal reason being that, in the process of neutralising the solution, some of the disulphonate must be decom- posed, yielding sulphate. The numbers fop the trisulphonate are much more satisfactory, being less than 1.7 per cent,. too high, apparently because they represent the difference between two sulphate determinations, the error in t h e one counterbalancing the correspond- ing error in the other, When, however, we come to the numbers for the disulphonate, which are calculated from those for the other substances, it is Been how large the error may become, being in one case as much as 11.4 per cent.too low. Similarly, the :quantity of nitrite, calculated from those of the other substances, may come out as much as 4.5 per cent. too high, The expression of the errors as percentages only holds good, of course, where the salts in an actual experiment are nearly in the same proportions as here taken, as they were generally found to be. The quantities of nitrite and of gases yielded by the peroxylamine- sulphonate may also be each determined directly. The nitrite may be estimated by the urea method, as stated on p. 94, most of the sulphonate salts having first been crystallised out and washed with alcohol. The method for collecting and measuring the gases produced during the decomposition consists in letting this proceed in a closed vessel from which the air is withdrawn.A stout-walled, cylindrical bolt-head, of about 250 C.C. capacity, with a stopcock sealed on to it, mas exhausted106 RAGA : PEROXYLAMINESULPHONATE8 AND and then opened with its mouth in the solution of peroxylamine- sulphonate and potassium hydroxide. About 200 C.C. were allowed t o enter, holding between 6 and 7 grams of the salt in solution. The tube was again exhausted and the stopcock being then closed, the salt was decomposed by heating the solution. When cold, the apparatus was connscted with a Sprengel pump and the gases drawn off and measured. They proved t o be free from nitric oxide, but, on treat- ment with strong alcohol, a small proportion of nitrogen remained un- dissolved.The experiments on this method of determining the gases have been very few, and not such as have admitted of their utilisation in this paper, beyond giving proof that nitrogen in small quantity is generated along with the nitrous oxide, which is the main constituent of the gaseous mixture, and that the quantity of the gases may vary greatly in different experiments. Where the decomposition of the peroxylaminesulphonate proceeds in the presence of lead peroxide, as it is made to do in the prepara- tion of hydroxylaminetrisulphonate, no hydroxylaminedisulphonate can remain in the solution, and in place of i t is found principally an increase in the quantities of trisulphonate and nitrite. The absence of the disulphonate simplifies the analysis, as is seen on p.94. Eight analyses of the products of the spontaneous decomposition of the peroxylaminesulphonate were made. In Expt. 1, a solution holding 2.31 84 grams of potassium peroxylaminesulphonake and 60 C.C. of N/lO solution of potassinm hydroxide was made up to 150 C.C. and then found to weigh 150.79 grams. It was slowly heated to the boil- ing point., and kept boiling till decolorised. Expt. 2.-The sollition, weighing 234.3 grams and containing 2.547 grams of the salt and 72.3 C.C. of N/10 potassium hydroxide, was left in the cold for two days, and then boiled until colourless. During the boiling, a reflux condenser was used to retain the water in the solution. Expt. 3.-A solution, weighing 120.42 grams and measuring 120 C.C. of 1.22 grams of salt and 11-4 C.C. of N/lO potassium hydroxide, was decomposed by boiling. Expt. 4.-The solution weighed 134.9 grams, and contained 1.8576 grams of salt and 30 C.C. of N/10 potassinm hydroxide; i t was decomposed by boiling. Expt. 5.-A solution of 0.6601 gram of salt and 80 C.C. of X/10 potassium hydroxide, weighing 34.25 grams, was left in the cold for a week, when it had become colourless. Expt. 6.-Like the last, but the solution weighed 266.7 grams and the salt 5.27 grams, whilst the potassium hydroxide was taken in about the same proportion as before. Expt. 7.-The solation was a portion of the same as had been used for Expt. 2. When kept in a closely- stoppered flask, it had only lost all its coloiir after about five months. Expt. &--This experiment differed from the others in the use of baryta-water in place of potassium hydroxide, and to this must beHYDROXY LAMINETRISULPHONATES. 107 attributed the production of so much sulphate and hydroxylamine- disulphonate. I n the table, the ncmbers of molecules of the several products yielded by lOO(SO,K),N,O, are given according t o calculation from the analyses made in the way above described, and without any corrections €or the "probable, but variable, errors inherent in the method. The solution took a month t o lose all its colour. RIol. xeight. 1. 2. 3. 4. 5. 6. '7. 8. (SO,K),NO ......... S5 102.3 101.7 77.7 91.4 55.3 86 12.9 (SO,,I<),NOII: ...... 61 30 23-6 65.4 42-3 42 40.4 133.2 S0,KH ............ 23 33 47.8 36 41.2 60 61.2 94.7 HNO, ............... 49 36 b2'5 50'2 35 37 26'4 - Without fuvther experiments, i t does not seem possible to account for the wide variations in these numbers, except where baryta was used. From Expts. 2 and 7, started on portions of the same solu- tion, i t seems clear that, with the slow decomposition of the peroxyl- aminesulphonate which goes on in the cold, instead of the rapid change which occurs at a boiling heat, molecular quantities of hydroxyl- aminedisulphonate and sulphate take the place of some of the hydroxylaminetrisulphonate, and t h a t a little of the hydroxylamine- disulphonate is replaced by its equivalent of sulphate and nitrous oxide. This becomes more obvious when equations in these two cases are given with only 12 molecules instead of 100 molecules of decomposing peroxylaminenulphonate. This is possible without deviating from the numbers forind more than t'he imperfections of the analytical method allow. (2) 12(SO,K),N,O, + 6ET,O = 12(SO,K),NO + 4(SO,K),NOH + (7) 12(S03K),N,01 + 8H,O = lO(SO$),NO + 5(SO,K),NOH + 4SO,KB + 4KO,H + 2N,O. 8SO,HH + 3NO,H + 3N,O. The production of small quantities of nitrogen aad aminemono- sulphonate is of necessity ignored in the above calculabions. The author gratefully acknowledges his indebtedness to Dr. Divers, F.R. S., for a thorough revision of his manuscript. COI.T,R:GF, OF SCIENCE, IMITRIAL UNIVERSITY OF TOKYO.