270 PROFESSORS KIRCHHOFF AND BUNSEN ON CHEMICAL XX1V.-ON CHEMICAL ANALYSIS BY SPECTRUM-OBSERVATIONS. RY PROFESSORS KIRCHHOFF AND BUNSEN. (From Poggendorff’s Annslen Bd. cx. S. 161.) is well known that many substances possess the power of developing peculiar bright lines in the spectrum of a flame into which they are introduced. This property may be made the foundation of a method of qualitative analysis which greatly enlarges the field of chemical reactions and leads to the solution of problems hitherto unapproachable. The Fright lines produced in this manner shorn themselves most plainly when the temperature of the flame is highest and its illuminating power least. Hence Bunsen’s gas-burner which gives a flame of very high temperature and very slight luminosity is well adapted for experiments on the bright lines of the flame- spectra produced as above described.The coloured plate at the end of this paper shows the spectra given by such a flame when the chemically pure chlorides of potassium sodium lithium strontium calcium and barium are it Comp. rend ,xxxvii 110. 9 Vol. vi p. 273. $ Ann Ch. Phgs. 131 xxxviii 38. AKALYSIS BY 8PECTRUM-OBSERVATIONS. volatilised in it. The ordinary solar spectrum is added by way of comparison. The potassium compound employed was obtained by igniting chlorate of potassium which had been previously recrystallized six or eight times. The chloride of sodium was prepared by neutralizing pure carbonate of sodium with hydrochloric acid and crystallizing the salt several times.The salt of lithium was purified by precipitation fourteen times with carbonate of ammonium. The purest specimen of' marble which could be obtained dissolved in hydrochloric acid was the source of the calcium salt. From this solution the carbonate of calcium was thrown down in two portions by fractional precipitation with carbonate of ammonium the latter half of the calcium salt being converted into the nitrate. The salt thus obtained was dissolved in absolute alcohol and after evaporation of the alcohol converted into the chloride by precipi- tation with carbonate of ammonium and solution in hydrochloric acid. To obtain pure chloride of barium the commercial salt was boiled out repeatedly with nearly absolute alcohol and the residual salt freed from alcohol was dissolved in water and thrown down by fractional precipitation in two portions.The second only of these portions was dissolved in hydrochloric acid and the salt still further purified by repeated crystallizations. Pure chloride of strontium was prepared by crystallizing the commercial salt several times from alcohol and by fractional pre- cipitation of the salt with carbonate of ammonium the second portion being dissolved in nitric acid and the nitrate freed from the last traces of calcium salt by boiling with alcohol. The pro-duct thus purified was lastly thrown down with carbonate of ammonium and the precipitate dissolved in hydrochloric acid. All these various purifications were conducted as much as possible in platinum vessels.The apparatus which we have usually cmployed for our spectrum-observations is represented in the annexed woodcut. A is it box blackened on the inside having its horizontal section in the form of a trapezium and resting on three feet; the two inclined sides of the box which are placed at an angle of about 58O from each other carry the two small telescopes B and C. The eye-piece of the first telescope is removed and in its place is inserted a plate in which a slit made by two brass knife-edges is so arranged that it coincides with the focus of the object-glass. The gas-larnp D stands before the slit in a position such that the mantle of the flame is in a straight line with the axis of the telescope B. Some-what lower than the point at which the axis of the tube produced meets the mantle the end of a fine platinum wire bent round to a hook is placed in the flame.The platinum wire is supported in 272 PROFESSORS KIRCH~TOFFAND BUNSEN ON CIZEMICAL this position by a small holder E and on to the hook is melted a globule of the dried chloride which it is required to examine. Between the object-glasses of the telescopes B and C is placed a hollow prism Fg filled with bisulphide of carbon and having a refracting angle of 60'. The prism rests upon a brass plate move- able about a vertical axis. The axis carries on its lower part the mirror G and above that the arm Hi which serves as a handle for turning the prism and mirror. A small telescope placed some way off is directed towards the mirror and through this telescope an image of a horizontal scale fixed at some distance from the mirror is observed.By turning the prism round every colour of the spectrum may be made to move past the vertical wire of the telescope C and any required position in the spectrum thus brought to coincide with this vertical line. Each particular portion of the spectrum thus corresponds to a certain point on the scale. If the luminosity of the spectrum is very small the wire of the telescope C may be illuminated by means of a leas which throws a portion of the ray3 fiom a Inmp throiigh a small opening in the side of the tube of the telescope C. We have compared the spectra represented 011 the Plate which were obtained from the pure chlorides with those produced when the bromides iodides hydrates sulphates and carbonates of the several metals are brought into the following flames :-Into the flame of sulphur., , bisulphide of carbon. , , aqueoiis alcohol. Into the non luminous flame of coal gas. Into the flame of carbonic oxide. , , hydrogen. Into the oxyhydrogen flame. As the result of these somewhat lengthy experimerits the details of which m7e here omit it appears that neither the alteration of ANA EY SIS BY SPECTRUM-ORSERVATIO his. the bodies with which the several metals may be combined nor the variety of the chemical processes occurring in the several flames nor the wide differences of temperature which these flames exhibit produce any efect upon the position of the bright lines in the spectrum which are characteristic of each metal.The following considerations show how much the temperature of these various flames differ. An approximation to the tempera- ture of a flame is obtained by help of the equation. t=Bgzu, ZPS in which t signifies the required temperature of the flame g the weight of one constituent of substance burning in oxygen ‘10 the heat of combustion of this constituent p the weight and s the specific heat of one of the products of combustion. The heat of combustion of the following bodies may be taken as-Sulphur ................. 224OOC. Bisulphide of carbon.. ..... 3400 Hydrogen.. .............. 34462 Marsh-gas.. .............. 13063 Ethylene................. 11640 Biitylene................. 11529 Carbonic oxide ........... 2403 The specific heats under constant pressure were fomd by Regnault to be- Sulphurous acid = 0.1553 Carbonic acid = 0,2164 Nitrogen = 0’2440 Aqueous vapour = 0’4750 Hence the temperatures of the flames are found to be- The sulphur flame.. .................. 1820°C. The bisulphiile of carbon flame. ........ 2195 The coal-gas flame*. .................. 2350 The carbonic oxide flame+. ............ 3042 The hydrogen flame in air$. ........... 3259 The oxyhydrogen flames .............. 8061 It was found ghat the same metallic compound placed in one of these flames gives a more intense spectrum the higher the tempe- rature of the flame. In the same flame those compounds of a metal give the brightest spectra which are most volatile.In order to prove stillmore conclusively that each of the above- mentioned metals always produces the same bright lines in the spectrum we have compared the spectra represented in the Plate with those produced when the electric spark passes between electrodes made of these metals. Small pieces of sodium potassium lithium strontium and calcium were fastened to fine platinum wires and melted two by * Ann. Ch. Pharm. cxi. 258. -f J3iinsen’s ‘Casometry.’ p. 242. $ Ibid. B bid. rr 274 PROFESSORS ItIRCEfHOFF AND BUXSlTh' ON CIIEJYIC'AL two into glass tubes so that the pieces of metal were separated by about 1to 2millims. and the platinum wires were melted through the sides of the glass tubes.Each of these tubes was placed in front of the spectrursn-instr.ument and by means of a Ruhn?korff's induction apparatus sparlrs were allowed to pass betwsen the pieces of metal inside the tube ; the spectritm thus produced v7as then compared with that given by a gas.fiame into which the chloride of the metal was brought. The flame was placed behind the glass tube. By alternately briaging the incluction apparatus into and out of action it was cmy without measuring to convince onaclves that in the brilliant spectrum of the electric spark the bright lines of the fiame-spectrum were present in their normal position. Besides these lines other bright ones appeared in the eIectric spark spectrum ; some of these were produced by foreign metals present in the electrodes ; others arose from nitrogen which filled the tubes after the oxygen had combined with a portion of the electrodes.From these facts it appears certain that the appearance of the bright lines represented in the spectra on the Plate may be regarded as absolute proof of the presence of the particular metal. They serve as reactions by means of which these bodies may be recognized with more certainty greater quickness and in far smaller quantities than can be done hy help of any other knom7n analytical method. The spectra drawn on the Plate represent those seen when tlie slit was of such a width that only tlie most conspicuous of the lines of the solar spectrum were visible tlie magnifying power of the telescope C being a fourfold one arid the light of a moderate degree of intensity.These circumstances seem to us to be the most advantageoils den it is requircd to make a chemical analysis by means of spectrum-observations. The appearance of the spec- trum may under other conditions be essentially different. If the purity of the spectrum be increased many of those lines which appeared before as single ones are split up into several; thus the sodium line is divided into two separate lines. If on the other hand the intensity of the light be increased new lines appear in several of the spectra and the relative brichtness of the old ones becomes altered. In general an indistiuct line becomes brigliter upon increasing the illumination more rapidly than it brighter line but not to such an extent that the indistinct line ever overtakes the brighter one in intensity.A good example of this is seen in the two lithium lines. We have observed only * On employing on one owasion with strontium-electrodes a tube filled with hydrogen instead of nitrogen the stream of sparks changed rapidly into a continuous arc of light whilst a grey pellicle covered the inside of the tube. The tube was opened under rock-oil when it was found to tie cmpty the hydrogen having disap- penred. This gas appears at thc enormously high temperature of the electric spark Lo hare dccornposeci the oxide of strontium which was not completely removed from the metal. ANALYSIS BY SrECTRUM-OBSERVATIONS. one exception to this rule namely in the line Ba 7 which by light of small intensity is scarcely visible whilst Ba y appears plainly but by light of greater intensity becomes more visible than the latter.We intend on a future occasion to examine this poiut in detail. We now proceed to describe the peculiarities of the several spectra tlie exact acquaintance with which is of practical import- ance and to point out the advantages which this new method of chemical analysis possesses over the older processes. Sodium. The spectrum-renction of sodium is the most delicate of all. The pellow line Na a the only one which appears in the sodium spectrum is coincident with Fraunhofer’s dark line D and is remarkable for its exactly defined form and its extraordinary brightness.If the temperature of the flame be very high and the quantity of the substance employed very large traces of a continuous spectrum are seen in the immediate neighbourhood of the line. In this case too the weaker lines produced by other bodies when near the sodium line are discerned with difficulty and are often not seen till the sodium reaction has almost subsided. The reaction is most visible in the sodium-salts of oxygen chlorine iodine bromine sulphuric acid and carbonic acid ; but it is always evident even in the silicates borates phosphates and other non- volatile salts. Swan* has already remarked upon the small quantity of sodium necessary to produce the yellow line. The following experiment shows that the chemist possesses no reaction which will bear the remotest comparison as regards delicacy with this spectrum-analytical determination of sodium.In a far corner of our experiment room the capacity of which was about 60 cubic metres we burnt a mixture of 3 milligrammes of chlorate of sodium with milk-sugar whilst the non-luminous flame of the lamp was observed through the slit of the telescope. Within a few minutes the flame which gradually became pale yellow gave a distinct sodium line lastiug for ten minutes and then entirely disappearing. From thc weight of sodium salt burnt and the capacity of the room it was easy to calculate that in one part by weight of air there was suspended less than 2+m of a part of soda-smoke. As the reaction can be quite easily observed in one second and as in this time the quantity of air which is heated to ignition by the flame is found from the rate of issue and froin the composition of the gases of the flame to be only about 50 cub.cent. or 0.0647 grm. of air containing less than ‘k0 of sodium salt it folloms that the eye is able to detect 4 * Trans. Roy. SOC.Edinb. vol. xxi. Part 111. p. 411. T2 276 PROFESSORS HIRCHHOFF AND BUNSEN ON CHENICAZ with the greatest ease quantities of sodium salt less than +m of a milligramme in weight. With a readion so delicate it is easy to understand why a sodium reaction is almost always noticed in ignited atmospheric air. More than two-thirds of the cart h’a surface is covered with a solution of chloride of sodinm fine particles of which are continually being carried into the air by the action of the waves.The particles of sea-water thus cast into the atmosphere evaporate leaving almost inconceivably small ivsidues which floating about are almost always present in the air and are rendered mident to our eyesight in the sunbeam. These minute particles perhaps serve to supply the smaller organ-ized bodies with the salts which larger animals and plants obtain from the ground; but there is also another point of view in which the presence of this chloride of sodium in the air is of interest. If, as is scarcely doubtful at the present time the explanation of the spread o€ contagious disease is to be sought for in some peculiar eontact-action it is possible that the presence of an antiseptic substance like chloridc of sodium eveii in almost infinitely small quantities may not be without influence upon such occurrences in the atmosphere.By means of daily and long-continued spectrum observations it would be easy to discover whether the alteration of intensity in the line Na a produced by the presence of sodium-compounds in the air has any connection with the appearance ad direction of march of an endemic disease. The unexampled delicacy of the sodium reaction explains also the well-observed fact that all bodies after a lengthened exposure to air show the sodium line when brought into a flame and that it is only in a few salts that it is possible to get rid of the last traces of the line Na a,even after repeated crystallization from water which has only been in contact with platinum.A thin platinum wire freed by ignition from every trace of sodium salt shows the reaction most visibly after a few hours’ exposure to the air. In the same way the dust which settles from the air in a room shows the bright line Na a to render this evident it is only necessary to knock a dnsty book for instance at a distance of some feet from the flame when a wonderfdly bright flash of the yellow band is seen. Lithium. The luminous ignited vapour of the lithium compounds gives two sharply defined lines the one a very weak yellow line Li 8 and the other a bright red line Li a. This reaction likewise exceeds in certainty and clelicacy all ordinary methods of ana- lytical research.It is however not quite so sensitive as the sodium reaction only perhaps because the eye is more adapted to distinguish yellow than red rays. When 9 milligrammes of carbonate of lithium mixed mit h excess of milk-sugar were burnt the reactioii Raq visible in a room of 60 cubic metres capacity. AN.\LYSIS BY STECTRU~i-OBSERVATIONS. Hence by the method already explained we find that the eye is capable of distinguishing with absolute certainty a quantity of carbonate of lithium less than +w of a milligramme in weight 0.05 grm. of carboi'ate of' lithiuin salt burnt in the same room was su6cient to enable the ipited air to show the red line Li a for art hour after the combustion had taken place. The compounds of lithium with oxygen iodine bromine and chlorine are the most suitable for the purpose; still the carbonake sulphate and even the phosphate give almost as distinct a reaction.Minerals containing lithium such as triphylline triphane petalite kpidolite require only to be held in the flame in order to obtain the bright lirie Li a in the most satisfactory manner. In this way the presence of lithium in many felspars can be directly shown as for instance in the orthoclase from Baveno. The lirie is seen for a few moments only directly after the mineral is brought into the flame. In the same way the mica from Altenberg and Penig was found to contain lithium whereas micas from Miask Ashaffen- burg Modum Bengal Pennsylvania &c. were found to be free from this metal.In natural silicates which contain only small traces of lithium this metal is not observed so readily. The examination is then best conducted as follows :-A small portion of the substance is dipted and evaporated with hydrofluoric acid or fluoride of ammonium the residue moistened with sulphuric acid and heated the dry mass being dissolved in absolute alcohol. Tlie alcoholic extract is then evaporated the dry mass again dissolved in alcohol and the extract allowed to evaporate on a shallow glass dish. The solid pellicle which remains is scraped off with a fine knife and brought into the flame upon the thin p€atinum wire. For one experiment &-of a milligramme is in general quite a sufficient quantity. Other compounds besides the silicates in which small traces of lithium require to be detected are transformed into sulphates by evaporation with sulphuric acid or otherwise and then treated in the manner described.In this way we arrive at the unexpected conclusion that lithium is most widely distributed throughout nature occurring in almost all bodies. Lithium was easily detected in 40 cubic centimetres of the water of the Atlantic Ocean collected in 41' 41' N. lati-tude and 39" 14' W. longitude. Ashes of marine plants (kelp) driven by the Gulf-stream on the Scotch coasts contain evident traces of this metal. All the orthoclase and quartz from the granite of the Odenwald which we have examincd contain lithium. A very pure spring water from the granite in Schlierbach OD the west side of the valley of the Neckar was found to contain lithium whereas tile water from the red sandstone which supplies the Heidelberg laboratory was shown to contain none of this metal Mineral waters in a litre of which lithium could hardly be detected by the ordinary methods of analysis gave plainly the line Li a even if only a drop of the water on a platinum wire was 278 PROFESSORS KIRUHHOFF AND BUNSEN ON CIIEMICAL brought into the flame.* All the ashes of plants growing in the Odenwald on a granite soil as well as Russian and other potashes contain lithium.It was found also in the ashes of'tabacco of vine leaves of the wood of the vine and of grapes,? as well as in the ashes of the crops grown in the Rhine-plain near Waghausel Deidesheim and Heidelberg on a non-granitic soil.The milk of the animals fed upon these crops also contains 1itfiium.f It is scarcely necessary to say that a mixture of volatile sodium and lithium salts gives the reaction of lithium alongside that of sodium with a scarcely perceptible diminution of precision and distinctness. The red lines of the former substance are still plainly seen wlien the bead contains part of lithium salt and when to the naked eye the yellow soda-flame appears untinged by the slightest trace of red. In cotisequence of the somewhat greater volatility of the lithium salt the sodium reaction lasts longer than that of the other metal. In those cases therefore in which small quantities of lithium have to be detected in presence of large quantities of sodium the bead must be brought into the flame whilst the observer is looking through the telescope.The lithium lines are often seen only for a few moments amongst the first products of the volatilization. In the production of lithium salts on the large scale in the proper choice of a raw material and in the arrangement of suitable methods of separation this spectrum-analysis affords most valuable aid Thus it is only iiecessary to place a drop of mother-liquor fmm any mineral spring in the flame and to observe the spectrum produced in order to show that in many of these waste products a rich and hitherto unheeded source of litliiurn salts exists. In the same way during the course of the preparation any loss of lithium in the collateral products and residues can be easily traced and thus more convenient and economical methods of preparation may be found to replace those at present employed.$ Potassiuna.Volatile potassium compounds give when placed in the flame a widely extended continuous spectrum which contains only two characteristic lines namely one line ICa a in the outermost red * When liquids have to be brought into the flame it is best to bend the end of the platinum wire of the thickness of a horsehair to a small ring and to beat this ring flat. If a drop of liquid be brought into this ring enough adheres to the wire for the experiment. j-In the manufactories of tartaric acid the mother-liquors contain so much fith'um salts that considerable quantities can thus be prepared.$ Dr. Folwarczny has been able by help of the line Li a to detect lithium in the ash of human blood and of muscular tissue. 3 Ke obtain by such an improved method from two jars (about 4 litres) of a mother-liquor from a mineral spring which by evaporation with sulpburic acid gave 1h-2 of residue half an ounce of carbonate of lithium of the purity of the commercial the cost of which is about 140 florins the pound. A great number of other mineral- spring mot!ier-liquors which we examined showed a similar richness in compounds of lithium. ANALYSIS BY SPECTRW M-OBSEBVATIONS. approaching the ultra-red rays exactly coinciding with the dark line A of the solar spectrum and a second line Ka p situated far in the vioiet rays towards the other end of the spectrum and also identical trritli a particular dark line observed by Fraunhofer.A very indistinct line coinciding with Fraunhofer’s line B which however is seen only when the light is very intense is not by any means so characteristic. The violet line is somewhat pale but can be used almost its well as the red line for the detection of potassium. Owing to the position of these two lines both situated near the limit at which our eyes cease to be sensitive to the rays this reaction for potassium is not so delicate as the reaction for the two metals already mentioned. It became visible in the air of our room when one granime of clilorate of potassium mixed with milk-sugar was burnt. In this way therefore the eye requires the presence of r$m of a milligramme of chlorate of potassium in order to detect the presence of potassium.Caustic potash and all compounds of potassium with volatile acids give the reaction without exception. Potash silicates and other non-volatile salts on the other hand do not produce the reaction by themselves unless .the metal is present in very consi- derable quantity ; when however the amount of potassium is smaller it is merely necessary to melt the substance with a bead of carbonate of sodium. The presence of the sodium does not in the least interfere with the reaction and scarcely diminishes its delicacy. Orthoclase sanidine and adularia say in this way be easily distinguished from albite oligoclase Labradorite and anorthite. In order to detect the smallest traces of potassium salt the silicate requires only to be slightly ignited with a large excess of fluoride of ammonium on a platinum capsule after which the residue is brought into the flame on a platinum wire.In this way it is found that almost every silicate contains potassium. Salts of lithium diminish or influence the reaction as little as sodium salts. Thus we need only to hold the end of a burnt cigar in the flame before the slit in order at once to see the yellow line of sodium and the two red lines of potassium and lithium this latter metal being scarcely ever absent in tobacco ash. Strontium. The spectra produced by the alkaline earths are by no means so simple as those produced by the alkalies. That of strontium is especially characterized by the absence of green bands.Eight lines in the strontium-spectrum are remarkable namely six red one orange and one blue line. The orange line Sr a whiclt appears close by the sodium line towards the red end of the spectrum the two red lines Sr and Sr y and lastly the blue line Sr 6 are the most important strontium bands both as regafds their position and their intensity. To examine the interlsity of 280 PBOFESSORS RIRCHHOFE AND SUX'SEN ON CHEMICAL the reaction we quickly heated an aqrieous solution of chloride of strontium of known concentration iu a platilium dish over a large flame till the water was evaporated and the basin became red-hot. The salt then began to decrepitate and was thrown up into the air in microscopic particles in the form of a white cloud.On weighing the residual salt it mas found that in this way 0.077 grm. of chloride of strontium had been mixed in the form of a fine dust with the air of the room weighing 77000 grms. As soon as the air in the room was perfectly mixed by rapidly moving an open urn brella the characteristic lines of the strontium-spectrum were beautifully seen. According to this experimeiit a quantity of strontium may be thus detected equal to the &m part of a milligramme in weight. The chloride and the other haloicl salts of strontium give the best reaction. The hydrate and carbonate of strontium give it much less vividly the sulphate still less whilst the com-pounds of strontium with the non-volatile acids give either a very slight reaction or none at all.Hence it is well first to bring the bead of substance alone into the flame and then again after moist- ening with hydrochloric acic?. If it be supposed that sulphuric acid is present in the bead it must be held in the reducing part of the flame before it is moistened with hydrochloric acid for the purpose of changing the sulphate into the sulphide which is decomposed by hydrochloric acid. To detect strontium when combined with silicic phosphoric horacic and other non-volatile acids it is best to proceed as follows:-Instead of fusing with carbonate of sodium in a platinum crucible a conical spiral of platinum wire is employed; this spiral is heated to whiteness in the flame and dipped while hot into finely-powdered dried carbonate of sodiutn which should contain so much water that a sufficient quantity adheres to the wire when it is once dipped into the salt.The fusion takes place in this spiral much more quickly than in a platinum crucible as the mass of platinum requiring heating is small and the flame comes into direct contact with the salt. As soon as the finely-powdered mineral has been brought into the fused soda by means of a small platinum spatula and the mass retained above the melting point for a few minutes the cooled mass has only to be turned upside down and hocked on thc porcelain plate of the lamp in order to obtain the salt in a large coherent head. The fused mass is covered with a piece of writing paper and then broken by pressing it with the blade >f a steel spatula until the whole is reduced to a fine powder.The powder is brought to one spot on the edge of the plate and care- fully covered with hot water which is allowed to flow backwards and forwards over it so that after decanting and rewashing the powder several times all the soluble salts are extracted without losing any of the residue. If a solution of chloride of sodium be employed instead of water the operation may be conducted mom rapidly and with greater security. The insoluble salt contains the strontium as carbonate; and one or two tenths of a milligramme of the substance brought on to the wire and moistened with hydrochloric acid is sufficient to produce the most intense reaction. It is thus possible without help of platinum crucible mortar evaporating basin or funnel and filter to fuse powder digest and wash out the substance in the space of a few minutes.The reactions of potassium and sodium are not influenced by the presence of strontium. Lithium also can be easily detected in presence of strontium when the proportion of the former metal is not very small The lithium line Li a appears as an intensely red sharply defined band upon a less distinct red ground of the broad strontium band Sr p. Calcium. The spectrum produced by calcium is immediately distiiiguislied from the four spectra already considered by the very characteristic bright green line Ca 6. A4second no less characteristic feature in the calcium spectrum is the intensely bright orange line Ca a,lying considerably nearer to the red end of the spectrum than either the sodium line Na a or the orange band of strontium Sr a.By burning a mixture consisting of chloride of calcium chlorate of potassium and milk-sugar a white cloud is obtained which gives the reaction with as great a degree of delicacy as strontium salts do under similar circumstances. In this way it was found that &m of a milligramme in weight of chloride of calciuni can be detected with certainty. Only the volatile compounds of calcium give this reaction; the more volatile the salt the more distinct and delicate does the reaction become. The chloride bromide and iodide of calcium are in this respect the best compounds. Sulphate of calcium does not produce the spectrum till it has become basic but then very brightly and continuously.In the same way the reaction of the carbonate becomes more distinctly visible after the carbonic acid has been expelled. Compounds of calcium with the non-volatile acids remain inac- tive in the flame; but if they are attacked by hydrochloric acid the reaction may easily be obtained as follows A few milligrammes of the finely powdered substance are placed on the moistened flat platinum ring in the moderately hot portion of the flame so that the powder is fritted but not melted on to the wire; if a drop of hydrochloric acid be now allowed to fall into the ring so that the greater part of the acid remains hanging to the wire and if the wire be then brought into the hottest part of the flame the drop evaporates in the spheroidal state without ebullition.If the spec- trum of the flame be observed during this operation it will be noticed that at the moment when the last particles of liquid evaporate a bright calcium spectrum appears. If the quailtitics of the metal present are very small the characteristic lines are 282 PROFESSORS KIRCHHOFF AND BUNSEN ON CHEXICAZ seen for a moment only; if larger quantities are contained the phenomenon lasts for a longer time. It is only in silicates which are decomposed by hydrochloric acid that the calcium can be thus detected. In those minerals which are not attacked by that acid the examination is best made as follows. A few milligrammes of the substance under examina- tion in a state of fine division are placed upon a flat platinum lid together with about a granme of fluoride of ammonium and the mixture is gently ignited until all the fluoride is volatilized.The slight crust of salt remaining is moistened with a few drops of sulphuric acid and the excess of acid removed by heat. If about a milligramme of the residual sulphates be scraped together vith a knife and brought into the flame the characteristic spectra of potassium sodium and lithium supposing these three metals to be present are first obtaiued either simultaneously or consecutively. If calcium and strontium be also present the corresponding spectra generally appear somewhat later after the potassium sodium and lithium have been volatilized.When only traces of strontium and calcium are present the reaction is not always seen; it becomes however immediately apparent on holding the bead for a few moments in the reducing flame then moistening it with hydro- chloric acid and again bringing it into the flame. These easy experiments such as either heating the specimen alone or after moistening with hydrochloric acid or after treating the powder with fluoride of ammonium either alone or in presence of sulphuric or hydrochloric acid provide the mineralogist and geologist with a series of most simple methods of recognizing the components of the smallest fragment of many substances (such for instance as the double silicates containing lime) with a cei tainty which is attained in an ordinary analysis only by a large expen- diture of time and material.The following examples will illustrate this statement. 1. A drop of sea-water heated on the platinum wire shows at first a strong sodium reaction ;and after volatilization of the chloride of sodium a weak calcium spectrum is observed which on moistening the wire with hydrochloric acid becomes at once very distinct. If ft few decigrammes of the residual salts obtained by the evaporatiou of sea-water be treated in the manner described under lithium with sulphuric acid and alcohol the potassium and lithium reac- tions are obtained. The presence of strontium in sea .water can be best detected in the boiler-crust from sea-going steamers. The filtered hydrochloric acid solut#ion of such a crust leaves on evapo- ration and subsequent treatment with a smsll quantity of alcohol a residue slightly yellow-coloured from basic iron salt which is deposited after some days and can then be collected on a small filter and Fvashed with alcohol.The filter burnt on a fine platinum wire and held in the flame gives besides the calcium lines an intensely bright strontium spectrum. ANALYSIS BY SPECTRUM-OBPERVATIONS. 283 2. Mineral waters often exhibit the reactions of potassium sodium lithium calcium and strontium by mere heating. If for example a drop of the Diirkheim or Kreuznach water be brought into the flame the lines Na a Li a Ca a and Ca p are at once seen. If instead of using the water itself a drop of the mother- liquor be taken these bands appear most vividly.As soon as the chlorides of sodium and lithium have been to a certain extent volatilized and the chloride of calcium has become more basic the chzracteristic lines of the strontium spectrum begin to show them- selves and continue to increase in distinctness until at last they come out in all their true brightness. In this case therefore by the mere observation of a single drop undergoing vaporization the complete analpis of a mixture containing five coustituents is per-fomned in a few seconds. 3. The ash of a cigar moistened with hydrochloric acid and held in :the flame shows at once the bands Na a Ka a Li a Ca a Ca p. 4. A piece of hard potash-glass combustion tubing gavk both with and without hydrochloric acid the lines Na a and Ks a; treated with fluoride of ammonium and sulphuric acid the bands Ca a Ca @ and traces of Li a were rendered visible.5. Orthoclase from Baveno gives either alone or when treated with hydrochloi+c acid only the line Na a with traces of Li a and Ka a; with fluoride of ammonium and sulphuric acid the bright lines Na a and Ka a and a somewhat less distinct Li a are seen. After volatilization of the bodies thus detected the bead moistened with hydrochloric acid gives a scarcely distinguishable flash of the lines Ca a and Ca p. The residue on the platinum wire when moistened with cobalt solution and heated gives the blue colour so characteristic of alumina. If the well-known reaction of silicic acid be likewise observed we may conclude from this examiuation made in the course of a very few minutes that the orthoclase from Baveno contains silicic acid alumina potash with traces of soda lime and lithia; and also that no trace of barrpta or strontia is present.6. Adularia from St. Gothard coniported itself in a similar manner excepting that the calcium reaction was indistinctly seen whilst that of lithium was altogether wanting. 7. Labradorite from St. Paul gives the sodium line Na a but no calcium spectrum On moistening the fragment with hydro- chloric acid the lines Ca a and Ca /3 appear very distinct; with the fluoride of ammonium test a weak potassium reaction is obtained and also faint indications of lithium. 8. Labradorite from the Corsican diorite gave similar reactions except that no lithium was found.9. Mosanderite from Brevig and Tscheff kinite from the Ilmengebirge showed when treated alone the sodium reactioll ; 011 the aclclition of hydrochloric acid the lines Ca a and Cs p. 284 PROFESSORS KIHCHHOFF AND BUNSEN ON CMEBLPCAL 10. Melinophane from Lamoe gave the line Na a when placed alone in the flame; with hydrochloric acid the lines Ca a Ca ,@ and Li a became visible. 11. Scheelite and sphene give on treatment with hydrochloric acid a very intense calcium reaction. 12. When small quantities of strontium are present together with calcium the line Sr 6 may be most conveniently employed for the detection of this metal. In this way the presence of small quantities of strontium may be easily detected in very many sedi- mentary limestones.The lines Na a Li a IKa a especia1.lp Li a are observed as soon as the limestone is braught into the flame. Converted by hydrochloric acid into chlorides and brought in this form into the flame these minerals give the same bands; and not unfrequently the line Sr 6 is also distinctly seen. This latter appears however only for a short time and is in general best seen when the calcium spectrum begins to fade. In this way the lines Na a Li a Ka a Ca a Ca @ and Sr 6 were fobnd in the spectra of the following limestones :-Limestone from the Silurian at Kugelbad near Prague. Mus-chelkalk from Rohrbach near Heidelberg. Limestone from the Lias at Malsch in Baden. Chalk from England. The following limestones gave the lines Na a Li a Ka a,Ca a Ca /3 but not the blue strontium band Sr 8 :-Marble from the granite near Auerbach.* Devonian limestone from Gerolstein in the Eifel.Carboniferous limestone from Planitz in Saxony. Dolomite from N ordhausen in the Ham. Jura-kalk from Streitberg in Franconia. From these few experiments it is evident that a more extended series of exact spectrum analyses respecting the amount of stron- tium lithium sodium and potassium which the various limestone formations contain must prove of the greatest geological import- ance both as regards the order of their formation and their local distribution and may possibly lead to the establishment of some unexpected conclusions respecting the nature of the oceans from which these limestones were originally deposited.Barium. The barium spectrum is the most complicated of the spectra of the alkalies and alkaline earths. It is at once distinguished from all the others by the green lines Ba a and Ba ,@ (which are by far the most distinct) appearing the first and continuing during the whole of the reaction. Ba ‘y is not quite so distiiict but is still a well-marked and peculiar line. As the barium spectrum is considerably more extended than those of the other metals the reaction is not observed to so grcat a degree of delicacy; still * According to the method already described a quantity of nitrate of strontium was obtained from 20 grms. of this marble such as to give a complete and vivid strontium spectrum. We have not examined the other limestones in the same way.ANALYSIS BY SPECTRUM-OBSERVATIOXS. 0.3 grm. of chlorate of barium burnt with milk-sugar gave a distinct band of Ba a which lasted for some time when the air of the room was well mixed by moving an open umbrella about. Hence we may calculate in the same manner as mas done in the sodium experiment that about &of a milligramme of barium salt may be detected with certainty. The chloride bromide iodide and fluoride of barium as also the hydrate the sulphate and carbonate show the reaction best. It may be obtained by simply heating any of these salts in the flame. Silicates containing barium which are decomposed by hydro- chloric acid also give the reaction if a drop of hydrochloric acid be added to them before they are brought into the flame.Baryta-harmotome treated in this way gives the lines Ca a and Ca /3, together with the bands Ba a and Ba p. Compounds of barium with fixed acids giving no reaction either when alone or after addition of hydrochloric acid should be fused with carbonate of sodium as described under strontium and the carbonate of barium thus obtained examined. If barium and strontium occur in small quantities together with large amounts of calcium the carbonates obtained by fusion are dissolved in nitric acid and the dried salt extracted with alcohol. The residue contains only barium and strontium both of which can almost always be detected. When we wish to test for small traces of strontium or barium the residual nitrates are converted into chlorides by ignition with sal-ammoniac and the chloride of strontium is extracted by alcohol.Unless the quantity of one or more of the bodies to be detected is extremely small the methods of separation just described are quite unnecessary as is seen from the following experiment :-A mixture of the chlorides of potassium sodium lithium calcium strontium and barium containing at the most -&-of a milligramme of each of these salts was brought into the flame and the spectra produced were observed. At first the bright yellow sodium line Na a appeared with a background formed by a nearly continuous pale spectrum. As soon as this line began to fade the exactly defined bright red line of Lithium Li a was seen; and beyond this still farther from the sodium line the faint red potassium line Ka a was noticed whilst the two barium lines Ba a Ra 6 with their peculiar shading became distkctly visible in their characteristic places.As the potassium sodium lithium and barium salts volatilized their spectra became fainter and fainter and their peculiar bands one after the other vanished until after the lapse of a few minutes the lines Ca a,Ca p Sr a Sr p Sr 7 and Sr 8 became gradually visible and like a dissolving view at last attained their characteristic distinctness colouring and position and then after some time bccsrrie pale and disap- peared entirely. 286 PROFESSORS KIRCHHOFF AND RUNSEN ON C;iImicIiT The absence of any one or of several of these bodies is at once indicated by the non-appearance of' the corresponding bright lines.Those who become acquainted with the various spectra by repeated observation do not need to have before them an exact measurement of the individual lines in order to be able to detect the presence of the various constituents ;the colour relative position peculiar form variety of shade and brightness of the bands are quite characteristic enough to ensure exact results even in the hands of persons unaccustomed to such work. These special distinctions may be compared with the differences of outward appearance presented by the various precipitates employed for detecting substances in the wet way. Just as a precipitate is characterized as gelatinous pulverulent flocculent granular or crys-.talline so the lines of the spectrum exbibit their peculiar aspects some appearing sharply defined at their edges others blended off either at one or both sides either similarly or dissimilarly some again appearing broader others narrower ;and just as in ordinary analysis we make use of those precipitates oiily which are produced with the smallest possible qiiantity of the substance supposed to be present so in analysis with the spectrum we employ only those lines which are produced by the smallest possible quantity of substance and require a moderately high temperature In these respects both analytical metbods stand on an equal footing; but analysis with the spectrum possesses a great advantage over all other methods inasmuch as the characteristic differ- ences of colour of the lines serve as the distinguishiag feature of the system.Most of the precipitates which are valuable as reactions are colourless ;and the tint of those which are coloured varies very consi-derably according to the state of division and mechanical arrangement of the particles. The presence of even the smallest quantity of impurity is often sufEcient entirely to destroy the characteristic colour of a precipitate; so that no reliance can be placed upon nice distinctions of colmr as an ordinary chemical test. In spectrum-analysis on the contrary the coloured bands are unaffected by such alteration of phy-sical conditions or by the presence of other bodies. The positions which the lines occupy in the spectrum indicate the existence of a chemical property as unalterable as the combining weights themselves and may Lherefore be estimat,ed with almost astronomical precision.The fact however which gives to this method of spectrum-analysis a peculiar degree of importance is that it extends almost to infinity the limits within which the chemical characteristics of matter have been hitherto confined. By an application of this method to geological inquiries concerning the distribution and arrangement of the components of the various fgrmntions the most valuable results nitiy be expected ;even the few random experiments already mentioned have led to the unex- pected conclusion that not only potassium and scdium but also lithium and strontium must be added to the list of bodies occurring only indeed in small quantities but most widely spread throughout the matter composing the solid body of our planet.The method of spectrum-analysis may also play a no less important part as a means of detecting new elementary substances; for if bodies should exist in nature SO sparingly difhsed that the analytical methods ANALYSIS BY SPECTRUM-OBSERVATIONS* 287 hitherto applicable have not succeeded in detecting or separating them it is very possible that their presence may be revealed by a simple exami- nation of the spectra produced by their flames. We have had opportunity of satisfying ourselves that in reality such unknown elements exist. We believe that relying upon unmistakeable results of the spectrum-analysis we are already justified in positively stating that besides potassium sodium and lithium the group of the alkaline metals contains a fourth member which gives a spectrum as simple and characteristic as thrtt of lithium-a metal which in our apparatus gives only two lines namely a faint blue one almost coincident with the strontium line Sr 6 and a second blue one lying a little further towards the violet end of the spectrum and rivalling the lithium line in brightness and distinctness of outline.The method of spectrum-analysis not only offers as we think we have shown a mode of detecting with the greatest simplicity the presence of the smallest traces of certain elements in terrestrial matter but it also opens out the investigation of an entirely untrodden field stretching far beyond the limits of the earth or even of our solar system.For in order to examine the composition of luminous gas we require according to this method only to see it; and it is evident that the same mode of analysis must be applicable to the atmospheres of the sun and of the brighter fixed stars. A modification must however be introduced on account of the light emitted by the solid nuclei of these heavenly bodies. In a Memoir published by one of us,* ‘(On the relation between the Coefficients of Emission and Absorption of Bodies for Heat and Light,” it was proved from theoretical considerations that the spectrum of an incandescent gas becomes reversed (that is that the bright lines become changed into dark ones) when a source of light of sufficient intensity giving a continuous spectrum is placed behind the luminous gas.From this we may couclude that the solar spectrum with its dark lines is nothing else than the reverse of the spectrum which the sun’s atmosphere alone would produce. Hence in order to effect the chemical analysis of the solar atmosphere all that we require is to discover those substances which when brought into the flame produce hright lines coinciding with the dark ones of the solar spectrum. In the paper above referred to the following experimental facts are given in confirmation of the preceding theoretical conclusion. The hright red line produced in the spectrum of a gas-flame by the presence of a bead of chloride of lithium is changed into a dark one when direct sunlight is allowed to pass through the flame.When the bead of lithium is replaced by one of chloride of sodium the dark double line D (coincident with the yellow sodium line) appears with uucomrnon dis- tinctness. The dark double line _O also appears when the rays of a Drummon d’s light are passed through the flame of aqueous alcohol into which chloride of sodium is thrown.+ * ITirchhofY Poggendorff’s Annalen cix. 275; and Phil. Mag. [4] XX. 1, + In the Pliilosopliical Magazine for March 1860 Prof. Stokes calls attention to the fact that in the year 1549 Foucault made an ohswvation very similar to hhe above. In the examination of the spectrum produced by tlie electric arc between carbon points Poucaul t noticed that bright lines occur where the double line D of the solar spectrum is found and that this dark line D is produced or made more intense when the rays of the sun or those froiii one of the incandescent carbon poles are passed through the himinous are.The observation mentioned in the text affords 288 KIRCHHOFF AND BUNSEN ON SPECTRUAI-ANBEYSL5. It appeared of interest to obtain still further confirmatioil of this important theoretical conclusion ; the following experiments answered this purpose :-We ignited a thick platinum wire in the flame and then by means of an electric current heated it to 8 temperature approaching its melting- point. The wire gave a bright spectrum in which no trace of either dark or bright lines was seen. A flame of weak aqueous alcohol in which common salt was dissolved on being brought between the wire and the slit of the apparatus gave the dark line 1') most distinctly.The dark line D can be produced in the spectrum of a platinum wire heated in a flame by holding between flame and spectrum a test-tube containing some sodium amalgam which is heated to boiling. This experiment is important because it shows that sodium vapour at a temperature much below that at which it becomes luminous exerts its absorptive power at exactly the same point of the spectrum as it does at the highest temperatures which we can produce or at the temperatures existing in the solar atmosphere. We have succeeded in reversing the bright lines in the spectra of K Sr Ca Ba by employing sunlight and mixtures of the chlorates of these metals with milk-sugar.A small iron trough was fixed in front of the dit of our apparatus in which the mixture was placed; the direct sun- light was then allowed to pass along the whole length of the trough and the mixture was igaited with a heated wire. The telescope C with the wires cutting each other at an oblique angle was placed so that the point of intersection of the wires coincided with the bright line of the flame- spectrum which was to be examined. The observer concentrated his attention upon this point to judge whether at the moment of burning the mixture a dark line showed itrelf passing through the point of intersection of the cross wires. In this way it was easy when the right proportions for the mixtures were found to show that the lines Ba a Ba p as well as the line Ka p were reversed.The last of these lines coincides with one of the most distinct dark lines in the solar spectrum though not marked by Fra un h o f er which however appears inuch more plainly than it is generally Reen at the moment the potash salt burns. In order to prove that the strontium lines can be reversed the chlorate of strontium must be most carefully dried as the slightest trace of' moisture produces a positive strontium spectrum owing to small particles of salt being thrown about in the flame and thus diminishing the power of the solar rays. an explanation of this interesting phaenomenon observed by F o 11 c au 1t eleven years ago proving that it is not occasioned by the properties of the electric light which in many respects is still so enigmatical but that it arises from a compound of sodium contained in the pole and converted into incandebcent gas by the current.