It will be readily understood from this explanation that tests made at the gas works, or even at points arbitrarily selected at a certain distance from these works, are hardly calculated to satisfy the consumer. For this reason I have preferred, in conducting these tests, to sacrifice to some degree the accuracy that obtains in laboratory experiments, in order to test gas samples _taken from the main directly in front of the complainant’s own premises_. I argue that the consumer cares little or nothing as to whether the gas as manufactured complies with the law, or whether tests made at a point perhaps a mile away from the works show the required candle power; but that he does want to know what is the quality of the gas passing in at his service pipe. The method of collecting and transporting to a laboratory the gas samples enables one to say with positiveness that the gas at the point of complaint has an illuminating power of at least so many candles, and that it may be even one candle better than the tests indicate. The figures thus obtained range from twenty and a half to twenty-five. So, then, the gas _delivered to the consumer_ is not ‘poor.’ Hygienic reasons demand that the impurities in the gas shall not exceed a definite percentage. Whatever effect these impurities may have upon the candle power has been covered by the tests above explained, so that any further consideration of these impurities may be omitted here. It is always a difficult matter to convince an indignant householder that the quality of the gas supplied to him is satisfactory. He knows perfectly well that he is not getting the desired result, and no explanation, however elaborate, as to candle power will placate him, unless it be supplemented by a further statement detailing the cause of the trouble. When you are trying to draw water in the bathroom while the cook is filling the washtubs in the basement, do you say the water is ‘poor’? Why, then, should you characterize the gas as ‘poor,’ when people nearer to the gas works than you are happen to be drawing heavily upon the common gas main? Imagine, if you please, a long gas main, with consumers tapping in at points throughout its entire length, and with a gas holder forcing the gas in at one end. Since there is a loss of pressure, caused by the transmission, it follows that the pressure will be higher at the gas holder than anywhere else along the line, the _difference_ in pressure depending, roughly, upon the size and length of the pipe and upon the amount of gas flowing. Now, for any one customer the size and length of pipe will remain constant, but the flow of gas along the line will vary from hour to hour, consequently the pressure at his house may be expected to vary from hour to hour.