GIFT OF GEOLOGICAL SURVEY OF PENNSYLVANIA. J. P. LESLEY, STATE GEOLOGIST. THE COMPOSITION AND FUEL VALUE NATURAL GAS. BY PROF. FRANCIS C. PHILLIPS. WESTERN UNIVERSITY, ALLEGHENY CITY. Extract from Annual Report for 1 886. HARRISBURG: EDWIN K. MEYERS, STATE PRINTER. 1887. The Chemical Composition of Natural Gas. BY FRANCIS C. PHILLIPS, Professor of Chemistry, Western University, Allegheny, Penn'a. Introduction. Natural gas, as obtained from several of the most produc- tive fields in Pennsylvania, according to the analytical data presented in this report, consists chiefly of the hydrocar- bons of the paraffin series, together with nitrogen, a small proportion of carbon dioxide and traces of t oxygen. Free hydrogen was found in minute quantity in Speechley gas. It is possible that by employing many thousand cubic feet of gas, traces of other constituents might be discovered. Inasmuch as the composition of natural gas possesses an interest for those who are not familiar with the strictly chemical aspect of the question, a few preliminary state- ments as to the more characteristic properties of its chief constituents will no doubt prove of value in this connection. Hydrogen is obtained as a gas by the action of dilate sulphuric acid upon zinc. It is also produced during the putrefaction of vegetable matters buried under stagnant water. Its specific gravity is 0.069234 as compared with NOTE. Prof. Phillips has spent considerable time in the study and prac- tical investigations of gaseous fuels, and at my request he was commissioned in the early part of the year to make analyses of the natural gas from eight of the most prominent pools in the State, and one analysis of the Fredonia gas in New York. The first systematic investigation as to the composition of natural gas in the State, was made by the Geological Survey in 1875, the results of which were published in a Report on the Use of Natural Gas in Iron Manufacture, in 1876. Since 1883, when the use of natural gas for fuel became more gen- eral, numerous analyses of the different gases have been made by a number of chemists. The wide differences in the composition of the gases as shown by these analyses were so great that Prof. Phillips exercised more than special care in the collection of his samples andln the method of determin- ing the individual constituents of the gases. All analyses were made in duplicate. C. A. ASHBURNER, Geologist in Charge. 2 GEOLOGICAL 'StmVv OF PENN'A, 1880. air. One cutic'nieter weighs >G. 089523 kilogram. One cubic foot weighs 39.12 grains. Hydrogen is odorless and taste- less. It takes fire at a bright red heat, and more readily than other constituents of fuel gases. Hydrogen in burning generates 34180 heat units per unit weight burned. The product of its combustion is water. In fuel gases hydrogen may occur in two very different forms. In its free or uncomhined state, it is often reported in the analyses of natural gas, and constitutes generally from 30 per cent, to 40 per cent, by volume of ordinary coal gas, being a product of the destructive distillation of coal at very high temperatures. The presence of a large proportion of free hydrogen in a gas fuel causes it to burn with a relatively small admixture of air, since one volume of hy- drogen requires only one-half volume of oxygen, or two and one-half volumes of air for complete combustion. The hydrogen flame is non-luminous. In combination with carbon, in the form of hydro-carbons, hydrogen constitutes about one-fourth by weight of the combustible portion of the natural gas now being used as fuel in Pennsylvania. These hydro- carbons, which represent approximately nine-tenths by volume of natural gas, are divided into two classes : Paraffins and Olefines. Of the paraffins, the best known and most abundant is methane (C'H 4 ) consisting of 25.03 per cent, hydrogen, and 74.97 per cent, carbon by weight. Methane is, like hydrogen, a product of the destructive distillation of coal, and consequently constitutes a large proportion of ordinary coal gas. It is also produced with hydrogen when plants decay at the bottom of rivers and swamps, and hence its older name of marsh gas. Methane, when pure is odorless, and not poisonous. Its specific gravity is 0.55297. One cubic meter weighs 0.7148 kilogram. One cubic foot weighs 312.36 grains. It is converted into a liquid under a pressure of about 2700 Ibs. per square inch at 12 F., or at 263 below zero F., under atmospheric pres- sure. Methane requires twice its volume of oxygen or ten PJttUipS.'] COMPOSITION OF NATURAL GAS. 3 volumes of air for its complete combustion, and the pro- ducts are carbon dioxide and water vapor. The Hukill well, Lyon's run, south of Murrysville, as already stated, yields this gas in a nearly pure condition. Methane contains in one cubic foot, two cubic feet of hy- drogen, and hence in the union of the carbon and hydrogen, a considerable condensation occurs. Methane is the typical and best known member of a large group of hydro-carbons, which exhibit a remarkable resemblance in chemical rela- tionships. The following list includes several of the most important : Methane, C H 4 Ethane, C 2 H 6 Propane, C 3 H 8 Butane, C 4 H 10 Pentane, C 5 H 12 Hexane, C 6 H 14 Heptane, C 7 H 16 Octane, C 8 H 18 C n H 2 n-f 2 The first four hydro-carbons are gases, but are more and more easily condensable to the liquid form in proportion as the amount of carbon is greater. The higher paraffins are solid. Common ''paraffin wax" contains several of the high- est members. While Methane (C H 4 ) constitutes from 50 per cent, to 90 per cent, or more of Pennsylvania natural gas, Ethane, (C 2 H 6 ), the next member of the series occurs in smaller quantity. Concerning the higher members, Pro- pane, (C 3 H 8 ), and Butane, (C 4 H 10 ), very little is as yet known, but there is reason to think that they are of com- mon occurrence. Pentane, (C 5 H 12 ), is found in the lightest distillates from petroleum, and the higher members are found in abundance in crude oil. It may be said concern- ing the gaseous hydro-carbons of the series that they pos- sess higher specific gravity, fuel value and illuminating power, and also stronger odor in proportion as the per- centage weight of carbon is greater. The illuminating power of pure methane, artificially pre- pared, has been determined as 5.15 to 5. 20 standard candles 4 GEOLOGICAL SURVEY OF PENN'A, 1886. per 5 cubic feet burned per hour. (Wright, Chemical News, 1885, p. 102.) The second class of hydro-carbons found in gas and petroleum includes the Oleh'nes. Of these the typical mem- ber is Ethylene or Olefiant gas, (C 2 H 4 ). Ethylene is one of the products of the action of heat upon coal and various vegetable substances. It is a gas having a specific gravity of 0.96744. Condensable to a liquid at a temperature of 166 below zero F. According to Frankland its illuminating power is equal to 68 standard candles, and hence the name 'illuminating hydro-carbons" often give to the group. One cubic foot in burning requires 3 cubic feet of oxygen, or 15 cubic feet of air. On account of their limited occur- rence, oleiines in many cases have no influence upon the fuel value of natural gas. They appear to be more abundant among the less volatile hydro-carbons of petroleum. Whether hvdroii'en occurs in the free state in a gas fuel, or as a hydro-carbon, the product of combustion will in- variably be water vapor, mixed in the latter case with car- bon dioxide. Carbon Dioxide, CO 2 . Well known as a universal pro- duct of decay, and as a gaseous furnace product, Carbon Dioxide, or Carbonic Acid is everywhere present, in the air, in water and in the soil and rocks. A suffocating gas, having a specific gravity of 1.5241. 1 cubic meter weighs 1.9650 kilogram. Condensable to a liquid under 780 Ibs. pressure at 60 F. Being incombustible its presence in gas (varying from a trace to 4 or 5 per cent.) tends to reduce to a corresponding degree the fuel value. Its presence may readily be shown by causing the gas to stream slowly through lime water, in which a milky deposit of carbonate of lime soon begins to form. Nitrogen. As a diluent of greater influence upon fuel value, we must regard nitrogen, on account of its occurrence in larger quantity. Constituting f of atmospheric air, it is \\ell known for its chemically indifferent character. In gas fuels it reduces the heating power in proportion to its ouantitv. Phillips.^ COMPOSITION OF NATUKAL GAS. 6 Gas from the Hnkill well, Lyon's run, contained 2.02 per cent, while gas from Houston (near Canonsburg) contained 15.30 per cent, of nitrogen. Should the natural gas supply ever become seriously diminished, it is probable that a time will come when the actual calorific power will be an im- portant factor in determining the market value. In that event the proportion of carbon dioxide and nitrogen, as well as the character of the hydro-carbons, will possess great in- terest for the gas companies and the consumers. Oxynen being well known as the constituent of atmos- pheric air which is the active cause in all cases of combustion slow or rapid, its presence in natural gas would seem im- probable. Contact of oxygen with the oxidizable elements of gas under high pressure would appear likely to cause its absorption and the formation of a corresponding amount of carbon dioxide or water. Nevertheless minute traces are constantly found and are indicated with great x>ositiveness in gas as it flows directly from the wells and under high pressure. It has been experimentally shown that oxygen and nitrogen may be dissolved and held in mechanical so- lution by petroleum, and that oxygen is even more soluble in petroleum than in water. (St. Guiewosz, Reports of the Berlin Chemical Society, 1887, p. 188.) For its liquifaction methane requires, as already stated, a pressure of at least 2,700 Ibs. at common temperatures. Ethane is liquified under a pressure of 690 Ibs. Carbon dioxide requires a pressure of 780 Ibs. Far greater pressures are needed for the liquifaction of oxygen, nitrogen and hydrogen. It is a fact of much interest in this connection that in the case of methane, the principal constituent of natural gas, the pressure under which liquii'action takes place is about four times that found in the most productive gas wells. It' in the reservoir tapped by the well a pressure exists four times greater than that at the well mouth, it is probable that the expansion there resulting would cause a marked lowering of the temperature in the well. It is commonly found however that the main leading from the well mouth does not possess a temperature much lower G GEOLOGICAL SURVEY OF than the air. From this it seems probable that methane cannot exist in a liquified state in the rocks. The carbon dioxide and ethane, on the other hand, may occur constantly in liquid form in the rocks to which many of the wells penetrate. Collection of Samples. G-lass vessels having a capacity of 250 to 400 cubic centi- meters were carefully dried by a current of warm air, and in order to obtain the gas as nearly as possible free from moisture the folio-wing method was employed : Glacial phosphoric acid, partially cooled from fusion, was drawn out into fine threads. A considerable number of such threads, in short pieces, could be pushed through the glass stopcocks, by which the vessels were closed, and left in the vessels which were then ready for the reception of gas samples. It is of importance to state that these vessels had been long in use for the same purpose arid had been proved to be air-tight by thorough and repeated tests. In collecting the samples several of these glass cylinders were connected in a series with the well or main by a short rubber hose, and gas allowed to flow for twenty minutes through them all. The stopcocks were then closed in such a manner as to leave a slight excess of gas pressure in each vessel. The stopcocks (which had previously been well greased with a mixture of tallow and wax) were then wound over and completely covered by fine cord, so that each resembled a ball of cord. The capillary ends of the cylinders were then closed by short pieces of thick rubber hose plugged with glass rods. By this mode of wrapping all movement of the stopcocks during transportation on railroads is prevented. The gas thus left in contact with the glacial phosphoric was gradually dried and ready for analysis on reaching the laboratory. The common method of taking a gas sample in a glass cylinder having finely drawn out ends, which are to be sealed by a flame when the vessel is filled, is not applicable l'hillipS.~} COMPOSITION OF NATURAL GAS. 7 in the case of natural gas. The constant escape of gas about a gas well renders the use of a liame absolutely im- possible on account of the danger of accident. Vessels closed by glass stopcocks are now supplied by dealers capable of holding a gas sample for many weeks without risk of leaking. Method of Analysis. The determination of carbon and hydrogen existing in combustible form in the gas was conducted by combustion over oxide of copper in a porcelain tube, which was kept at a bright red heat, and the resulting carbon dioxide and water collected separately and w'eighed. One of the glass cylinders, tilled with gas at the well, was placed in a vertical position and the temperature observed at intervals. When it was found that the temperature had remained constant for two hours, the lower stopcock was opened for a moment to allow the excess of gas to escape and secure equilibrium between the pressure of the gas inside and that of the atmosphere. At the same time the temperature and the height of the barometer were recorded. The glass cylin- der was then connected with a porcelain tube containing oxide of copper, and already heated to intense redness in a furnace, and the gas forced out of the cylinder by dry mer- cury. As the gas escaped from the cylinder it was carried through the porcelain tube by a slow stream of nitrogen previously dried by suitable means. The gas was thus burned completely to carbon dioxide and water which were collected and weighed by the usual methods, using a balance plainly sensitive to -I^^Q-Q gram. After the combustion, the glass cylinder was accurately calibrated by means of mercury at a known temperature, and thus was determined the exact volume of gas which had been burned. As it appeared possible under the conditions of the method that some nitrogen might undergo an oxidation, the water produced in the combustion of the gas was care- fully tested, but in no case was the water found to have an acid reaction. 8 GEOLOGICAL SURVEY OF PENN ? A, 1886. In the above described method are determined the weights of carbon and hydrogen per unit volume of gas. In con- ducting the combustion great care was taken to secure complete oxidation of the combustible constituents, and absorption of the products. For the absorption of the water sulphuric acid of 1.71 Sp. Gr., followed by phosphoric anhydride, was used, and for the carbon dioxide a solution of caustic potash in glycerine. For the determination of nitrogen the following method was employed : A porcelain combustion tube containing oxide of copper was brought to a yellow heat, and a stream of carbon dioxide conducted through the tube until the last traces of air were expelled. The expulsion of the air was considered complete when it was found that the carbon dioxide escaping from the tube was wholly absorbed by a solution of caustic potash. 100 cubic centimeters of such g.is not leaving a visible quantity unabsorbed by the alkaline solution. Then, after expulsion of the last traces of air, a quantity of natural gas (100 c. c. were generally employed), was allowed to flow slowly into the stream of carbon dioxide as it entered the combustion tube. In this manner the gas was burned and a mixture of nitrogen and carbon dioxide collected in a eudiometer over caustic potash solution. After the absorption of the car- bon dioxide the volume of the residual nitrogen was meas- ured. This nitrogen was carefully tested for carbon dioxide, oxygen and carbon monoxide, and was frequently repassed through the heated combustion tube a second time and again measured, in order to insure the complete combustion of all hydrocarbons. This repetition demonstrated in all but one or two instances that the nitrogen was pure. It was found that with a sufficiently slow stream of gas the oxidation by the oxide of copper is easily rendered com- plete, although the rate of flow must be regulated with great care. By the common eudiometric methods of analysis no de- termination is more difficult than that of nitrogen when occuring in small quantities in admixture with hydrocar- bons of the paraffin series. In the method above described COMPOSITION OF NATURAL GAS. 9 large quantities of gas can be employed, and the results are accurate. The determination of free oxygen in natural gas cannot well be made with the quantity of gas commonly at dis- posal. A test was made in every instance in about 100 cubic centimeters of gas, using an Elliott apparatus, and as an absorbent a solution of caustic soda and pyrogallic acid. In all cases the results were negative. I have found it necessary to conduct the tests for oxygen at the wells, and this was done in the following manner: A slow stream of gas was caused to flow (directly from the well or main) successively through solutions of caustic potash and pyrogallic acid, for 10 minutes, in order to expel dissolved air. Then by a simple contrivance the two fluids were mixed without interrupting the current of gas, which continued some time longer through the mixture. If the. mixed fluids then exhibited a brown color, gradually in- creasing in depth, it was considered that the presence of oxygen was established. The direct determination of free hydrogen has generally been considered a matter of such difficulty, that in many published analyses its quantity has been estimated by a calculation based upon the total carbon and hydrogen con- tained in the gas. For the present purpose a direct deter- mination seemed very desirable and the process of Hem pel has been used in the manner described below. 100 cubic centimeters of gas, after the removal of carbon dioxide, were washed with strong alcohol until the higher hydrocarbons, ethane, propane, &c. were removed. This was carried out in an Elliott apparatus, having a water jacket. Then the residual gas mixed with two or three times its volume of air was passed over asbestos coated with 30% of Palladium sponge at a temperature of 90C. By this treatment the hydrogen alone is burned, provided the higher jyaraffins, including ethane are previously re- moved by washing with alcohol. From the contraction in volume^after passing the palladium, the proportion of free hydrogen is easily determined. The method is very accurate when methane is the only 10 GEOLOGICAL SURVEY OF PEXN'A, 1886. hydro-carbon present. It is inaccurate in presence of ethane and the higher members of the series, and when these are present the washing with alcohol must be long continued. As it is a matter of great difficulty to retain hydrogen, even by the help of the most carefully ground stopcocks, the tests for this element were made in all cases at once after the arrival of the samples in the laboratory. The oleiines, as a group and carbon monoxide, are much more easily determined in natural gas than the paraffins and free hydrogen. The oletines are quickly absorbed and removed by bro- mine water and carbon monoxide by a solution of cuprous chloride. These reagents are used in the order named. Unfortunataly, however, these fluids are likewise solveuts, in less degree, for the paraffins, ethane, propane &c. Hence a gas perfectly free from olefines and carbon monox- ide is liable, on being washed with the above named fluids, to undergo a reduction in volume, leading to a wrong con- clusion. For the determination of these substances the following process was used, based on the solubility of both in a cuprous chloride solution. At the gas well a stream of gas was caused to bubble for two hours or more through 100 cubic centimeters of a solution of cuprous chloride. The solution was preserved for examination in the laboratory. A quart flask, provided with a gas delivery tube and a funnel tube reaching to the bottom, was filled with boiled water and then the cuprous chloride, prepared as above described, was poured into the flask through the funnel tube. The flask was then heated to the boiling point and the water caused to boil for three hours. A small quantity of gas was invariably collected from the cuprous chloride solution by this treatment. The gas so collected was transferred to an Elliott appa- ratus and carefully tested for olefines and carbon monox- ide by bromine water and cuprous chloride solution. In this way the quantities of these two constituents in a very large quantity of gas could be collected in concentrated form, convenient for a qualitative test. Phillips.'} COMPOSITION OF NATURAL GAS. 11 Carbon dioxide was determined by means of moist potash in a eudiometer over mercury, and also in the Elliott appa- ratus over water, by caustic potash solution. The latter method yields very correct results. In addition to the determinations carried out in the lab- oratory, the gas at the well was caused to pass in a slow stream through lime water. The stream of gas was made approximately the same by using the same delivery tube, depth of lime water and shape of containing vessel, and by counting the number of bubbles per minute, and then noting the rapidity with which the lime water became milky. For the detection of ammonia the gas at the well was caused to bubble through 100 c. c. of water, which had been carefully purified by distilling with addition of sul- phuric acid and permanganate of potash. This water was afterwards tested by Nessler's solution, after the common method in use in the examination of drinking water, for ammonia. The presence of exceedingly minute traces of ammonia could thus be shown with great accuracy. As solid masses of ammonium carbonate are reported to have been thrown out from the pipes leading from gas wells in the Hurry s- ville field tliis test seemed very important. In the statement of the results of analyses all gas vol- umes are to be understood as " normal,'' that is the vol- umes observed under different conditions of temperature and pressure are all reduced to zero, Centigrade, and 760 millimeters mercury pressure ; and, where measured in a moist condition, are calculated as dry. The temperatures were all measured by one and the same thermometer, of which the error was known from a com- parison with the Yale Observatory standard. This ther- mometer was made by Green in New York and is divided to T V degrees centigrade. The barometer used was made by Hicks, and indicated by vernier, changes of TTr l o inch. The constant error of this barometer was ascertained by -comparison with the standard barometer of the Signal Service department, in Washington. 12 GEOLOGICAL SURVEY OF PENN^A, 1886. In all cases of gas measurements in eudiometers, the observations were made by means of a Grunow cathetome- ter, having a millimeter scale and vernier and reading easily to ^ millimeter. The etched scales upon the eudiometer tubes, as com- monly supplied, are often very incorrect, both as regards uniformity and total length of scale, and are unsuited for accurate measurements of pressures or volumes. The glass cylinders containing the gas samples for com- bustion were calibrated at a temperature not differing by one degree Centigrade from the temperature at which the gas was measured for analysis. In this way the calculation of errors due to expansion and contraction of the glass vessels was rendered unnecessary. This necessitated re- peated calibrations after nearly ever} r combustion. In the calculation of the results of analyses, the follow- ing plan was adopted : The percentage of Carbon dioxide was determined volu- metrically. Having failed to find Carbon monoxide and oleh'nes in any of the samples, they are necessarily lel'r out of account in the calculation. Having found five hydrogen in only one of the gas samples, and herein traces, (Speechley,) it is also to be ignored in the calculations. The quantities of carbon dioxide and water produced in the combustion of a known volume of gas were weighed. From the weight of the water the proportion of hydrogen in a unit volume of gas could then be calculated. The pc r- centage volume of carbon dioxide contained in the gas being known, its weight was deducted from the weight of the total quantity obtained in the combustion. The dif- ference is the quantity corresponding to carbon in the form of hydrocarbons. The nitrogen having been determined in a separate portion of gas, and the free hydrogen being also known, the volume of the hydrocarbons will be expressed by the following equation Cs. Very little salt water is produced. The gas exhibits a de- cided oxygen reaction, turns lime water rapidly milky, and has a strong odor. Pipe lines carry the gas from these wells 2 PHILLIPS. 18 GEOLOGICAL SURVEY OF PENN'A, 1886. to Bradford, Jamestown, N. Y. ; Hornellsville, Salamanca, Buffalo, but the supply is largely in excess of the demand at present. Determination of (1.) (2.) Mean. Nitrogen, 9.32 9.50 9.41 per cent. Carbon dioxide, 0.21 0.20 0.21 per cent. Results of Analysis of Wilcox Gas. Nitrogen, .... Carbon dioxide, . Oxygen, Carbon monoxide, Olefines, Ammonia, .... Hydrogen, .... Paraffins, .... 100.00 374.2 cubic centimeters of Wilcox gas yield on combus- tion. H 2 O. 0.6022 gm, corresponding to H, 0.06706 gm=23.48 per cent. C O 2 .0.8014 gm, corresponding to C, 0.21856 gm=76.52 per cent. 100.00 Hence 1 liter paraffins contains : 0.64622 gm carbon. 0.19828 gm hydrogen. In the case of the Wilcox gas, an accident to some of the sample vessels prevented a second combustion, so that bat a single result can be presented. No. 4. Kane well, No. 1, at Kane, McKean Co. Gas col- lected Jan. 30th, 1887. The well was drilled in 1884. The pressure then was 550 fts when shut in for 40 minutes. It was allowed to blow off for 8 months, and then shut in, when the pressure increased to 630 Tbs. This gain in pressure has been permanent, up to October, 1886, when the last test was made. The Kane Natural Gas Co. own two other wells in addition to this. The gas exhibits decided oxygen and carbon dioxide reac- tions. Determination of (1.) (2.) Mean. Nitrogen, 9.67 9.91 9.79 Carbon dioxide, .... 0.20 0.20 0.20 rJt.tlUpS.\ COMPOSITION OF NATURAL GAS. 19 Results of Analysis of Kane Gas. Nitrogen, . . . . Carbon dioxide, . Oxygen, . . . . . Olefines, Carbon monoxide, Hydrogen, . . . . Ammonia, . . . . Paraffins, 100.00 349.03 cubic centimeters of gas yield on combustion. H 2 O, 0.5600 gm, corresponding to H, 0.06230 gm=23.18 per cent. C O 2 0.7580 gm, " " C, 0.20672 gm=76. 82 per cent, 100.00 Hence 1 liter of the paraffins contains : 0.65801 gm carbon. 0.19849 gm hydrogen. 248.1 cubic centimeters of the same gas yield on com- bustion. H 2 O, 0.3987 gm, corresponing to H, 0.04439 gm=23.28per cent. C O 2 , 0.5366 gm, " " C, 0.14634 gm=76. 72 per cent 100.W Hence the paraffins of Kane gas contain per liter: 0.19883 gm hydrogen. 0.65537 gm carbon. The means of these two analysis are per liter of paraffins, 0.65669 gm carbon = 76.77 per cent. 0. 19866 gm hydrogen=23.23 per cent. 1WKOO No. 5. Speechley. This field has been a remarkably productive one, as regards quantity and pressure of gas and number of wells. These wells are situated on a N. E. & S. W. line about 6 miles S. E. from Oil City. The sand rock from which the gas is obtained averages 1900 feet in depth, and is about 900 feet below the third oil sand of Venango county. This latter sand also produces gas, but in much smaller quantity, and it is consequently cased off, so that the gas in this territory is wholly obtained from one and the same 20 GEOLOGICAL SURVEY OF PENH 5 A, 1886. sand rock. The Northwestern Gas Co. of Oil City have 60 wells, and a large number of wells are owned by other com- panies. The samples of gas for examination were taken April 13th, 1887, from the high pressure main at South Oil City, belonging to the Northwestern Natural Gas Co. At this time the pressure in the main was 100 ibs. This sample may be considered to represent approximately the average of the gas from a large number of wells. ' The tests at the main indicated the presence of oxygen, but less of carbon dioxide than found in the Warren and McKean county gas. Determination of. (1) (2) Mean. Nitrogen, 4.61 4.41 4.51 per cent. Carbon dioxide, 0.05 0.05 0.05 Hydrogen, 0.02 0.02 0.02 Results of Analysis of Speechley Gas. Nitogen, 4.51 per cent. Carbon dioxide, 0.05 Hydrogen, 0.02 Carbon Monoxide, Olefins, Oxygen, trace. Ammonia, Paraffins, 95.42 100.00 304.24 cubic centimeters Speechley gas yield on combus- tion. H 2 O, 0.5423 gm, corresponding to H, 0.06039 gm=22.93 per cent. CO 2 , 0.7441 gm, " " C, 0.20293 gm=77.07 per cent 100.00 Hence the paraffins of this gas contain per liter 0,69900 gm Carbon 0,20801 gm Hydrogen In a second combustion of the same gas, 310.52 cubic centemeters yield H 2 O, 0,5500 gm, corresponding to H, 0.06125 gm=22.85 per cent. CO 2 ,0.7585 gm " " C, 0.20686 gm=77. 15 per cent. 100.00 P7lilHpS.~] COMPOSITION OF NATURAL GAS. 21 Hence the paraffins contain per liter : 0.20671 gm Hydrogen. 0.69815 gm Carbon. In a second combustion, 306.28 cubic centimeters of gas yield H 5 O, 0.4818 gm, corresponding to H, 0.05365 gm=25.02 per cent. CO 2 , 05895 gm " " C, 0.16074 gm==74. 98 per cent. 100.00 The means of these two results are per liter of paraffins : 0.69857 gm carbon = 77. 11 per cent. 0.20736 gm hydrogen = 22.89 per cent. 100.00 No. 6. Hukill well, on the Dick Farm, Lyons Run Dis- trict, southern end of Murrysville field, and one of 60 wells belonging to the Philadelphia Company. This well was drilled in 1883 and was allowed to blow off for a long time. The well is very productive and has a pressure as it flows through the main of 285 fbs. The well has extra heavy casing and there is good reason to suppose that the gas comes exclusively from the Mur- rysville sand. The sample was taken April, 8, 1887. The gas produces a decided carbon dioxide reaction but exhibits a very slight reaction for oxygen. This gas has a very faint odor, free from the pungent character noticed among some of the gas samples. The well yields no oil, but a very little salt water. Determinations of, (1) (2) Mean. Nitrogen, 2.13 1.91 2.02 Carbon dioxide, 0.26 0.30 0.28 Results of Analysis of Murrysville Gas. Nitrogen, 2.02 percent. Carbon dioxide, 0.28 Oxygen, trace Oarbon monoxide, Olefines, Ammonia, Hydrogen, Paraffins, 97.70 100.00 22 GEOLOGICAL SURVEY OF PENN'A, 188(5. 346,94 cubic centimeters of Murray sville gas yielded on combustion. H 2 O, 0.5473 gm, corresponding to H, 0.06095 gm= 25.06 per cent. CO 2 , 0.6682 gm, C, 0,18224 gm= 74.94 per cent. 100.00 Hence the paraffins in Murrysville gas contain per liter : 0,53763 gm Carbon. 0. 17981 gm Hydrogen. In a second combustion 306.28 cubic centimeters of gas yield H X O, 0.4818 gm, corresponding to H, 0.05363 gm=25.02 per cent. CO 2 0.5895 gm, " " C, 0. 16074 gm=74.98 per cent 100.00 .Hence the paraffins contain per liter : 0.53718 gm Carbon. 0. 17922 gm Hydrogen. The means of the above analyses are per liter of paraffins : 0.53741 gm Carbon = 74.96 per cent. 0.17950 gm Hydrogen = 25.04 per cent 100.00 The following experiments were tried at the valve house of the Philadelphia Company, in the rear of the office build- ing on Penn street, Pittsburg, beginning on March 22d, 1887. A Woulfe's bottle containing 200 c. c. purified water, and a second bottle containing cuprous chloride were con- nected with a gas meter, and gas allowed to stream slowly through them until 190 cubic feet had passed. The gas thus used comes directly from the Murrysville field. The gas was passed very slowly, so that 3 days were occupied in the transmission of the volume above named. The water was then tested for ammonia by Nessler's reagent No trace could be detected, although as is well known this reagent is capable of detecting ^nnnhrinro P art f ammonia in water, with great certainty. The cuprous chloride was tested for both olefines and Phillips."] COMPOSITION OF NATURAL GAS. 23 carbon monoxide by the method I have detailed, but no trace could be detected of either. The composition of methane gas by weight is Carbon, 74.97 per cent. Hydrogen, ...... 25.03 per cent. 100.00 Hence this Hukill well produces gas approximating in composition to pure methane, and in this respect differs from all those from which samples have been taken. It may be here stated that at the time the sample was collected there was every reason to believe thaf the gas came exclu- sively from this one well. No. 7. Raccoon Creek District. The sample was taken May 2d, 1887, from the high-pres- sure main of the Bridge water Natural Gas Co. at Rochester, Pa. The pressure at the time was 67 Ibs. The gas is produced wholly from one sand, which is about 1200 feet below the surface on Raccoon Creek, in Beaver county. The Bridgewater Company owns 23 wells and supplies the towns of Beaver Falls, Rochester, New Brighton, Phillipsburg, Vanport, Bridgewater, New Shef- field, Shannopin. The Youngstown Company own 12 wells in the same region. The gas is almost odorless, and the wells produce little or no salt water, and no oil. On causing the gas to bubble through lime water for 20 minutes the fluid remained perfectly clear. After 40 min- utes a rapid stream of gas caused the lime water to become faintly milky, as seen in a bright light. The proportion of carbon dioxide was far too small to allow of an accurate eudiometric determination. The oxygen reaction was faint but decided. This gas on being passed for one hour into a nitrate of silver solution produced a faint but decided reaction, indi- cating a trace of sulphuretted hydrogen. In the statement below, the result of the carbon dioxide test at the main is given. Determination of (1) (2) Mean. Nitrogen, ,. 10.00 9.82 9.91 24 GEOLOGICAL SURVEY OF PENN^A, 1886. Results of Analysis of Raccoon Greek Gas. Nitrogen, , . . . 9.91 per cent. Hydrogen, Carbon dioxide, trace. Carbon monoxide Olefines, Oxygen, trace Ammonia, Sulphuretted hydrogen, trace Paraffins, 90.09 100.00 In a combustion of Raccoon creek gas 325.48 cubic centi- meters yielded : H 2 O, 0.5108 gm, corresponding to H, 0.05688 gm= 23.60 per cent CO 2 , 0.6755 gm, " " C, 0.18422 gm= 76. 40 per cent. lOthOO Hence the paraffins in this gas contain per liter: . 0.62827 gm carbon. 0. 19398 gm hydrogen. In a second combustion 398.08 cubic centimeters gas yielded H 2 O, 0.6254 gm, corresponding to H, 0.06964 gm= 23.56 per cent. C O 2 ', 0.8286 gm, corresponding to C, 0.22598 gm= 76.44 per cent. 100.00 Hence the paraffins contain per liter: 0.63010 gm carbon. 0.19418 gm hydrogen. The means of these two results are per liter paraffins: 0.62918 gm carbon =76.42 per cent 0.19408 gm hydrogen= 23.58 per cent 100.00 This is the only gas which contains traces of sulphuretted hydrogen among those I have examined. No. 8. Baden, six miles S. E. from Rochester on the Pittsburgh, Fort Wayne and Chicago R. R., Beaver county. The samples were taken May 18th, 1887, from the Bryan well, No. 2, one of the four wells belonging to the Baden Gas Co. The gas is produced wholly from one sand which is 1396 feet deep, or about 1300 feet below the Ohio river. This well was drilled in May, 1886. Phillips.^ COMPOSITION OF NATURAL GAS. 25 The Baden wells are on the same anticlinal axis as the Kaccoon creek wells. This same axis continues northward a few miles east of the Speechley wells near Oil City. The gas exhibits a decided carbon dioxide and also an oxygen reaction. Determinations of (1.) (2.) Mean. Nitrogen, 12.26 22.38 12.32 per cent. Carbon dioxide, 0.41 0.41 0.41 Results of Analysis of Baden Gas. Nitrogen, 12.32 per cent Carbon dioxide, 0.41 Oxygen, trace Hydrogen, Carbon monoxide, Olefines, Ammonia, Paraffins, ... 87.27 100.00 317.17 cubic centimeters of Baden gas yield on combus- tion : H 8 O, 0.4892 gm, corresponding to H, 0.05447 gm= 23.48 per cent C O 2 , 0.6510 gm, corresponding to C, 0.17754 gm= 76.52 per cent 100.00 Hence the paraffins of Baden gas contain per liter : 0.64142 gm carbon. 0.19681 gm hydrogen. In a second combustion 332.70 cubic centimeters yield : H 2 O, 0.5130 gm, corresponding to H, 0.05712 gm= 23. 56 per cent CO 5 , 0.6843 gm, corresponding to C, 0. 18663 gm= 76. 44 percent 100.00 Hence the paraffins contain per liter : 0.64276 gm carbon. 0.19674 gm hydrogen. The means of these two results are per liter paraffins : 0.64209 gm carbon = 76.48 per cent 0.19677 gm hydrogen= 23.52 per cent 100.00 No. 9. Houston well, Houston station, 2 miles south of 26 GEOLOGICAL SURVEY OF PENN^A, 1886. Cannonsburg, on the Pittsburgh, Cincinnati and St. Louis. R. R., Washington county. This well is situated mile west of the station on Plum run. It is drilled nearly through the Gantz sand and is 1794 feet deep. An upper, gas producing, sand is found at 8,50 feet, but this is cased off so that the well may be considered to yield gas from the Gantz sand exclusively. The gas from the upper sand is said by well superintend- ants to burn with a whiter but more sooty flame than that from the greater depth. According to the statements generally heard at the wells, the occurrence of an upper, less productive gas sand, yield- ing gas of greater illuminating power, is a very common feature in many gas fields. The sample was collected on March 18, 1887. The gas exhibits an oxygen reaction and causes a rapid precipitation in lime water. Determination of (L) (2.) Mean. Nitrogen, 15.23 15.37 15. 30 per cent Carbon dioxide, 0.42 0.46 0.44 Results of Analysis of Houston Gas. Nitrogen, 15.30 percent. Carbon, dioxide 0.44 Oxygen, trace defines, Carbon monoxide, Ammonia, trace Hydrogen, Paraffins, 84.26 100.00 310.20 cubic centimeters of Houston gas yielded on com- bustion. H 2 O, 0.4601 gm, corresponding to H, 0.05124 gm,= 23.20 per cent. C O 3 0.6217 gm, " " " C, 0.16955 gm,= 76.80 per cent 100.00 Hence the paraffins contain per liter : 0.64871 gm carbon. 0. 19602 gm hydrogen. Phillips.'] COMPOSITION OF NATUKAL GAS. 27 In a second combustion 293.35 cubic centimeters yielded : H 2 O, 0.4392 gm, corresponding to H, 0.04891 gm= 23.44 per cent. C O 2 0.5855 gm, " " " C, 0.15968 gm= 76.56 per cent. 100.00 Hence the paraffins contain per liter : 0.64604 gm carbon. 0.19786 gm hydrogen. The means of these two analyses are per liter of paraffins : 0.64737 gm carbon = 76.68 per cent. 0.19694 gm hydrogen^ 23.32 per cent. 100.00 The analyses above detailed were carried out with great care, and every known precaution observed in order to se- cure accuracy. The results represent the character of the gas from par- ticular wells or groups of wells, scattered over a large region, and as it flowed from the wells on a single day. It is questionable whether they can be considered to rep- resent the average composition of natural gas, for the reason that the gas territory is so vast in extent. According to the above results natural gas is not so com- plex a substance as has been heretofore supposed. The samples examined may be said to consist mainly of the hydro-carbons of the paraffin series, among which methane predominates. It is to these bodies that the fuel value of the gas is due. Inasmuchas most of the gas conveyed through^pipe lines, deposits little or no liquid hydro-carbons, it is evident that the higher paraffins are not present in notable quantity. The method I have used in testing for the hydro-carbons of the olefine series enables me to state with much confi- dence that these bodies, ethylene, propylene, butylene, etc., are absent. Hydrogen I have found in Speechley gas alone, although the utmost care has been taken in the ex- amination. Perhaps still smaller quantities may have escaped detec- tion in other gas samples. 28 GEOLOGICAL SUKVEY OF PENN'A, 1886. Sulphuretted hydrogen was found only in Raccoon creek gas, but in faint traces. Oxygen is present in all, but in such small quantities that I have never succeeded in accurately determining its real percentage. As nearly as I can estimate, the Wilcox contains more oxygen than any other, and Murrysville the least. Ammonia was found, in traces only, in Houston gas. Carbonic oxide was not found in any of the samples. A comparison of the results in the accompanying table shows that the different gas samples differ mainly in the following particulars. 1. The proportion of carbon to hydrogen in the con- tained paraffins that is to say the ratio of the lower to the higher paraffins. Fredonia is seen to be the richest gas in carbon. 2. The proportion of nitrogen, which varies between 2.02% and 15.30%. The three gas fields, Speechley, Baden and Raccoon Creek approximately on the same anticlinal (according to Mr. I. C. White) produce gas having very different quantities of nitrogen. The resemblance between the Fredonia, Sheffield, Kane, Wilcox, and Raccoon Creek gas as regards the proportion of nitrogen is a matter of interest, although not explain- able. In the case of Murrysville, Speechley and Fredonia gas the density, richness in carbon, and calorific power of the contained paraffins are inversely as the proportion of nitro- gen. It is a curious fact that there is a certain continuity as regards composition in the case of the Fredonia, Kane, Sheffield and Wilcox gases, which disappears on reaching the Speechley field, in proceeding southward. South of Speechley much greater differences occur. 3. The carbon dioxide, which varies within very narrow limits. The only gas in which it almost disappears is that from Raccoon creek although Speechley gas contains barely more than a trace. Phillips.^ COMPOSITION OF NATURAL GAS. uxxjsnoH 8 t QO 8 jT?9u ' CM o o o a; o o CM t^ (NO S I-H O C^ O O O (M O O O O ^ ^* o d FH rH O O 0) O QO * (N O CO ^^ I 88 Oi O O O O O i-H 1- o CO O oid 2 8 so" ' !?> fl o y* (^) c &, 1 1 O tJD S| Eo 0) X5 ^J^ 1 B Carbon, . Hydrogen, 30 3EOLOGICAL SURVEY OF PENN'A 1886. At Oil City a sand is found 582 feet^ below low-water mark in the Allegheny river, which produces gas of lower pressure, amounting, it is said, to 20 ft>s. when shut in for some time. This gas is used in the Oil Well Supply Go's works for heating purposes. It bears the same relation to the Speechley gas sand 1900 feet deep as the shallow gas sands usually to the deeper, and more productive sand rocks. A determination of the nitrogen in the gas from this upper rock gave 5.62 per cent. Speechley gas contains 4.51 per cent. The sample was collected on April 13th, the day on which the Speechley samples were taken. The Speechley gas wells are six miles distant from this well. Tests for hydrogen, olefines, carbon monoxide and dioxide and ammonia in this gas all led to negative re- sults. Calculation of the Fuel Value of Natural Gas. The calorific power of any combustible may be determined by measuring the number of kilograms of water heated from to 1 C. by 1 kilo of the fuel' in burning, or by a cal- culation. The difficulties and inconveniences encountered in the first method necessitate commonly a resort to the second. Pure charcoal in burning produces, according to the re- searches of Favre & Silbermann (in 1849), 8080 heat units, or 1 kilo in burning will raise the temperature of 8080 kilos of water from to 1 C. By the same authors it was found that 1 kilo of hydrogen in burning generates a quantity of heat sufficient to warm 34462 kilos of water from to 1C that is 34462 heat units. Later determinations have been made by various authors, the most important being by Thomsen, who found 34180 (Berichte der Deutschen chemischen Gesellschaft, 1873, p 1533), and by Berthelot who obtained the number 34600, (Comptes Rendus, 1880 p 1240). The value assigned by Thomsen, viz : 34180, is probably the more correct. If it were possible that a fuel should contain pure hydro- gen and charcoal, uncombined, a calculation of its heating PMllipS.'] COMPOSITION OF NATURAL GAS. 31 power would lead to very correct results. It is found, however, that when a compound of carbon and hydrogen is burned, the number of heat units produced will not equal the number obtained when the same quantities of carbon and hydrogen are burned separately. Thus a kilo of methane produces 13270.5 heat units, but if the same quantities of carbon (as charcoal) and hydrogen were burned separately in a calorimeter, 14613 heat units result (assuming that the carbon produces 8080, and the hydrogen 34180 heat units per kilo burned). The difference between the calculated amount of heat, and the actually available heat 14613 13270=1313 heat units is 9.19 per cent, of the theoretical yield. For practi- cal applications this is a loss of heat, which must be con- sidered to represent the quantity of energy required to overcome the mutual affinity of the carbon and hydrogen, which are to be first separated, before they are burned to carbon dioxide and water. With more complex compounds the available heat of combustion does not fall so far short of the theoretical maximum, and it may be stated in a general way that the greater the number of carbon atoms in the compound, the more closely will the available and actual number of heat units coincide. This statement is especially true of cer- tain series of hydrocarbons. The following table (II) will serve to illustrate this in the case of the first three members of the paraffin series. For the higher paraffins no determina- tions have yet been made. GEOLOGICAL SUEVEY OF PENN' A, 1886. sjran jo TimtmxBin uo eyqBiTBAB jo IigjBJBd JO O\IJ[ J8J ns'^ara pauiui XI paujnq GJB pue jo inninTXBUi oq; aonpojd uaSoj'p^q pu uoqjBO aq^ ^Bq^ Suiinns -SB 's^mn !}B8q 25 2 Phillips.] COMPOSITION OF NATURAL GAS. 33 It lias been shown by Thomseri that isomeric hydro carbons, or those which differ in properties, although hav- ing identical composition, may produce different quantities of heat when burned, thus : Symbol. Heat Units. Propylene, C 3 H 6 11757 Trimethylene, C 3 H 6 10917 Difference = 840 The chemical formulas given show them to have the same composition, and yet these hydrocarbons would be rep- resented by different values if used as fuels. The presence of isomers among the hydro carbons of natural gas would tend to interfere with the correctness of a calculation of its fuel value. No isomers are known in the case of methane (CH 4 ). Berthelot has stated that a second hydro carbon isomeric with ethane (C 2 H 6 ) exists, which produces on burning 12776 heat units, instead of 12373, the number as deter- mined by Thorn sen. Thomsen's researches have disproved this assertion, how- ever, and have shown conclusively that ethane produced in a variety of ways invariably possesses the same calorific power. (Berichte der Deutschen chemischen Gesellschaft 1881, p 500). Isomers of the higher paraffins no doubt occur in gas, as well as in petroleum, but when it is con- sidered that in gas the higher paraffins occur only in small quantity, and moreover that the calculated and the avail- able calorific power differ much less in these higher mem- bers than in methane and ethane, the danger of error from the presence of such isomers cannot be considered likely to affect the calculated results. The calorific power of methane was determined by Andrews in 1848 as 13108 heat units (Philosophical Maga- zine 1848 p 321), and by Favre and Silbermann in 1853 as 13063. heat units. In 1880 Thomsen assigned it the value 13345.6, and this number agrees closely with that obtained by Berthelot in the same year, viz: 33343.8. More recently Thomsen has corrected his former result, and now gives 13270.5 as the 3 PHILLIPS. 34 GEOLOGICAL SURVEY OF PENN 5 A, 1886. most probable number. (Berthelot, Comptes Rendus, 1880 p 1240. Tliomsen, Berichte der Deutschen Chemise/hen Gesellschaft 1880, p 959 and 1321 Eef, and 1886 p 77, Ref.). The elaborate researches of Julius Tliomsen in thermo- chemistry, (Thermochemische Untersuchangen, Leipzig) have reached the fourth of a series of large volumes, and although designed primarily as a contribution to theoreti- cal chemistry, they supply data likely to prove of great value in the study of fuels for metallurgical and other technical purposes. The actual calorific power of a gas fuel may now, by the use of such data, be more satisfactorily determined by calculation, provided its composition is known, than by the use of a calorimeter. In this respect there is an im- portant difference between gas fuels and the various kinds of coal. Coal being a compound of carbon, hydrogen and oxygen, of a highly complex chacacter, or possibly a mix- ture of such compounds, no such plainly definable relation- ship exists between the theoretical maximum and the available heat quantity per unit weight burnt. The percentage composition by weight of the paraffins likely to occur in natural gas is expressed in the following table. Small quantities of condensible vapors of the higher paraffins occur in the gas in some places as is evidenced by the condensation of benzene in pipes. These heavier vapors occur usually in very minute quantity, if at all : III. Showing the Composition by weight of some of the Lower Paraffins. NAME. Symbol. Per cent, carbon. Per cent, hydrogen. Methane, CH 4 74.97 25.03 Ethane C 2 H 6 79.96 20.04 Propane, . . . C 3 H 8 81.78 18.22 Butane, C 4 H 10 82.72 17.28 Pentane, C 5 H 12 83.29 16.71 The analyses of natural gas above detailed show a varia- tion in the proportion of carbon and hydrogen in the case of the two extremes of 3.18 per cent., thus : Phillips. .] COMPOSITION OF NATURAL GAS. 35 The paraffins in Murrysville gas contain Carbon, 74.96 per cent, by weight. Hydrogen, 25.04 " " " 100.00 And in the case of Fredonia gas Carbon, 78.14 per cent, by weight. Hydrogen, 21.86 " " " 100.00 From the tabular statement of the composition of the lower paraffins, it appears that Murrysville gas, as ob- tained at the Hukill well, has nearly the composition of methane, while disregarding again the nitrogen and carbon dioxide present, the Fredonia gas, the richest in carbon, approximates in composition to a mixture of equal volumes of methane and ethane, of which the actual composition would be, by weight : Carbon, 78.22 per cent. Hydrogen, 21.78 per cent. 100.00 By this I do not imply that it actually contains these two paraffins in the proportion named, for it is possible that the gas in question contains more of methane and a very small quantity of some one of the higher paraffins, pro- pane, or quartane, etc! As I have stated in regard to the analyses, the exact de- termination of the percentage of individual paraffins is a matter of such extreme difficulty, that it may be considered practically impossible. If we assume that Fredonia gas really contains equal volumes of methane and ethane, and calculate its calorific power accordingly, the following error may be committed. The gas may contain a larger amount of methane than was assumed, and consequently a very small quantity of quar- tane or pentane, for although the percentage of carbon and hydrogen is definitely fixed ~by the analysis, it is still a question as to the arrangement of the carbon and hydro- gen in the form of higher or lower paraffins. 36 GEOLOGICAL SURVEY OF PENN'A, 1886. As the difference between the available and the theoreti- cal heat of combustion is greater in the case of methane and less and less in the higher paraffins, an under estimate of the quantity of methane would lead to too high a value for the available heat of combustion. On the other hand, an under estimate of the proportion of the higher para- ffinsj would cause the available heat, as expressed in heat units, to be rated too low, supposing that in both cases the absolute quantities of carbon and hydrogen remain con- stantly the same. This error would be small in most instances, but in the extreme case of 2 gases consisting of methane and ethane respectively, the error from this source would exceed 1%. I have attempted to correct this error, as will be shown be- low. The curious and intimate relationships of the paraf- fins are well illustrated by the fact that a mixture of 1 cubic meter each of methane, ethane and propane will con- tain the same proportions of carbon and hydrogen, and will consequently yield the same quantities on burning of C O 2 and H 3 O as three cubic meters of the intermediate hydro-carbon, ethane, 1 cubic meter of methane weighs 0.7148 kilo, and generates heat units, 9485 1 cubic meter of ethane weighs 1.34016 kilo, and generates heat units, 16582 1 cubic meter propane weighs 1.9656 kilos, and generates heat units, 23688 49755 ?ubic meters of ethane generate on burning heat units, . 49746 The numbers expressing the heat produced are obtained by multiplying the weight of the cubic meter by 18270, 12373 and 12052, respectively, as given in table II. The difference is so slight amounting to only 9 heat units, that it is evident that it would have been sufficiently accurate to assume this mixture of three hydro-carbons to consist of the intermediate member Ethane in so far as the calculation of the fuel value is concerned. Or it may be more broadly stated that, with a view to Phillips.} COMPOSITION OF NATUKAL GAS. 37 the calculation of the calorific power of natural gas, it is sufficiently accurate to assume that a natural gas, (contain- ing no hydro-carbons of the olefine series) has the simplest constitution consistent with its percentage by weight of carbon and hydrogen, and then to determine its fuel value accordingly. Fredonia gas, as shown in the table of analyses, consists of 90.05% of paraffins, together with 9.54% nitrogen and 0.41% carbon dioxide. The paraffins consist of 0.80423 kilo carbon and 0.22494 kilo hydrogen per cubic meter. The theoretical maximum of heat units for these paraf- fins is calculated as follows, per cubic meter : 0.80406X8080, 6497 0.22494X34180, . 7288 13785 When C II 4 burns, only 90.81% of the theoretical heat is available. When C 3 H 6 burns 92.95% can be utilized. Hence if Fredonia gas is to be looked upon as a mixture of equal volumes of the two hydro-carbons methane and ethane, it will contain about 1 and 1.87 parts by weight re- spectively, (or approximately two parts by weight) of methane and ethane. The available heat of combustion can be determined by multiplying the theoretical maximum by a factor which is intermediate between ^-foV- and -^f-^oS and as a very close approximation the fraction 2 Et -|- Mt 3 X 100 will, I think, be sufficiently accurate. In this Et = the percentage of available on theoretical maximum heat, for ethane and Mt = the same ratio for Methane. Substituting in this fraction 2 X 0.9295 -f 0.9081 =. 9224. 3 The theoretical maximum heat of combustion of the Fredonia gas, as calculated above, is 13785 heat units per cubic meter of contained paraffins. Then 13785x0.922412715 as the available heat units due to the paraffins in the gas. As there are 90.05% of paraffins, 38 GEOLOGICAL SURVEY OF PENT*' A, 1886. the remainder, consisting of nitrogen and carbon dioxide, the above number will be still further reduced, and 12715x0.9005=11450,= the available heat produced by 1 cubic meter of Fredonia gas. In the case of the gas from Sheffield, Kane, Wilcox, Rac- coon Creek, Baden and Houston, there is a general similarity as regards the percentage of carbon and hydrogen. Wilcox gas may be regarded as representing approximately the average, and as a calculation shows that a mixture of 4 volumes methane and 1 volume ethane contains carbon 76.54 and hydrogen. 23.46, we may, for the purpose of the present calculation, assume that the above mentioned six gases contain approximately these proportions of the two named paraffins. For such a mixture the factor by which to obtain the available calorific value will be 2 Mt 4- Et =0.9153. 3 X 100 This factor has accordingly been used in the case of the above named gases. Speechley gas may be considered to contain 5 volumes of Methane and 2 volumes of Ethane for the purpose of the present calculation, and the factor will be 3 Et -f 4 Mt =0. 9173. 7 X 100 Murrysville gas contains nearly pure methane, and con- sequently the factor will be 90.81. It is not implied in the above considerations that the actual proportions of what may be regarded as the most commonly occurring paraffins, CH 4 , C 3 H 6 , C 3 H 8 , etc., can be accurately stated, for this I believe to be impossible. These proportions have been assumed as not inconsistent with the analytical data, merely for the purpose of obtain- ing an approximately correct value for the factor to be used in the calculation of the calorific power of the gas. The following table (IV) contains the results of the calculations carried out as explained. Column No. 2 in this table ex- presses the quantities of carbon and hydrogen contained in 1 cubic meter of the paraffins in each gas. In column No. 3 are given the factors, the derivation and use of which have already been pointed out : PTiUUpS. ~] COMPOSITION OF NATURAL GAS. 39 s^S jo laoj oiqno OOT oi ^JJ9 Sup-tjaq ui jBnba psoo -j^qo ojnd jo spimoT'QjKi