22 THE COLLIERY GUARDIAN. July 3, 1914 RELATION OF FALLS OF ROOF IN COLLIERIES ON THE MIDDELBURG COALFIELD TO WEATHER CHANCES/ By Chas. J. Gray, A.R.S.M., F.G.S., M.I.M.E. It is well known that falls of roof kill more men than any other form of colliery accident. They can be prevented by a lavish use of timber, but their causes must be fully realised before such falls can be guarded against most economically. In the year 1912 the accident rate from falls of roof in the Middelburg coalfield increased, despite apparent greater care in working. The writer noticed a concen- tration of fatalities and accidents in the five months June to October, nine out of 11 fatalities and 10 out of 15 accidents occurring in those months. That the explanation was not to be found in fluctuations either in the numbers employed or in output per employee was shown by investigation of the labour and output returns for the collieries at which the accidents occurred. The question then arose as to whether there was any connec- tion between the increased number of accidents, concentrated in the dry months, and the unusually prolonged dry season. In order to decide that question, two lines of investigation were followed :— (1) Comparison between the annual accident rates and the rainfall. (2) Comparison between the number of accidents in particular months and the rainfall in those months. Following the first line it was found that for Witbank, the most central position on the coal field, rainfall records were available for the seven years 1906 to 1912 only, and that comparison must therefore be limited to that period. The comparison which works out as follows, is shown graphically in diagram 1. It will be noticed that the rainfall rose and fell in alternate years, rising three times and falling three times. The death- rate also rose thj:ee times and fell three times, but four out of the six changes were in a contrary direction to the changes of rainfall. The accident-rate rose twice and fell four times and five out of the six changes were Fig. 1.—Relation of Accidents and Death Rates to Rainfall. 10 Deaf, ■a. /90f> '90? . !9Ob '903 WO contrary to changes of rainfall. Taking the years in which the death rate and accident rate rose, and summing the changes, we find that + 3'51 death rate corresponds with — 0’64 rainfall, and 4- 2'67 accident rate with — 1'25 rainfall. Similarly for the years in which the rates fell, we find that —2’16 death rate corresponds with 4- 0’48 rainfall, and — 2'91 accident rate with 4- 1*09 rainfall. To partly eliminate the precaution factor, it seems, however, useful, and not unreasonable as a crude approximation, to assume that uniform increase in precaution since 1908, when both the accident and death rates reached their maximum, accounts for the difference between the 1908 and 1912 rates. That assumes an improvement of 0'31 per annum in the death-rates and 0'49 per annum in the accident rates. Summing up the changes as was done with the unadjusted rates, we find + 3’82 death-rate corresponds with — 0’64 rainfall - 1’23 „ „ „ + 0’48 + 3’53 accident „ „ „ — 2’52 „ - 1’80 „ „ „ „ + 2’36 The crude correction has made the correspondence between a change in the accident and death rates and a change in the reverse direction in the rainfall, still more marked than it was. From the foregoing it appears very highly probable that a decrease in the annual rainfall at Witbank corresponds with an increase in liability to accident from falls of roof in the collieries, and vice versa. Proceeding now to compare the numbers of accidents from falls of roof in particular months with the rainfall in those months, we find that for the seven years 1906 to 1912, the figures were as shown in diagram 2. * From a paper read before the Chemical, Metallurgical and Mining Society of South Africa. As would be expected with such small numbers of acci- dents, there is irregular fluctuation from month to month. A general tendency for the numbers to be high in the dry months is, however, noticeable. Thus of the five months in which the number of deaths was above the average 5*3, four are months in which the rainfall was below the average 2'51, and of the eight months in which the number of accidents was above the average 8'75, five are months in which the rainfall was below the average. The effect of fortuitous fluctuations can be lessened by taking longer periods than one month for comparison. The figures taken quarterly become:— Average Deaths. Accidents. monthly January to March ... 12 . ... 21 . rainfall. In. .. 4’68 April to June 18 . .. 30 . .. 0’76 July to September ... 18 . ... 30 . .. 0’38 October to December 16 . ... 24 . .. 4’21 Taken half-yearly and putting the six dry months together, we have:— Deaths. Accidents. Average monthly October to March 28 . .. 45 . rainfall. In. 4’44 April to September ... 36 . .. 60 . .. 0’57 A more direct comparison between accident and rainfall statistics is given in the following figures for the years 1906 to 1912:— Number of months in which rainfall was Less than 1 in.... 38 More than 1 in... 46 Accidents. . 58 .. . 47 .. Accidents per month. 1'52 1'02 Accident statistics are available for a longer period than rainfall statistics, and they have been examined from 1903 to see whether the concentration of accidents irvrr /O 9 * ^3* il 2 J 4 J 6 7 8 Fig, 2.—Relation of Accidents in Particular Months to Rainfall, Temperature and Humidity—(1906-1912 Inclusive). in the dry months is consistent from year to year. The figures show that deaths were most numerous in the dry months in five years and most numerous in the wet months in three years. Accidents were most numerous in the dry months in seven years and in the wet months in two years. The comparisons show that liability to accident from falls of roof is almost certainly much greater in the dry winter months than in the summer months. Apparently it is nearly 50 per cent, greater. Both lines of investigation have now been followed, and it has been found that accidents from falls of roof on the coal field have increased and decreased both annually and seasonally as the rainfall decreased and increased. That being so, it is morally certain that accidents and rainfall are connected. Eithei’ rainfall affects accident statistics (for it is unimaginable that accidents affect rainfall), or the same seasonal phenomena affect both accidents and rainfall. On consideration it seems that the latter must be the case, for percolation through the strata, which is the only known way in which the rainfall can intervene, would increase and not reduce the accident rate in the wet summer months. Though percolation is known to cause a few accidents, and may partly account for the number in December and January being higher than in November and February, its influence must be slight. Rejecting rainfall as the direct cause of the fluctuation in accident rates, we proceed to consider changes in temperature and air humidity, which are closely connected with rainfall. As complete records of those changes at Witbank are not available, the Middelburg records have to be used. Taking temperature first, the monthly comparison is shown graphically in diagram 2. It will be seen that the annual changes of mean temperature are very slight, and that there is no marked correspondence between them and changes in accident rates. Three times accident rates and mean temperatures changed in the same direction, and three times in a reverse direc- tion. A total decrease of 2'91 in accident rates corre- sponded with an increase of 0'61 in mean temperature, and a total increase of 2'67 in accident rates with an increase of 0 08 in mean temperature. In the monthly comparison we see a tendency to correspondence between concentration of accidents and low temperature, which is more clearly shown by the following figures:— Mean Accidents. Tempera- ture. October to March............... 45 ... 67T April to September............. 60 ... 55'3 Now, taking humidity as the basis of comparison, the monthly comparison is also shown graphically in diagram 2. In the annual figures, correspondence between increase of accidents and decrease of humidity, and vice versa, is almost as marked as the correspondence in the case of rainfall. Four times the changes were in reverse directions. A decrease of 2'91 in accident rates coincided with an increase of 13'6 in mean humidity and an increase of 2*67 in accident rates, with a decrease of 13'0 in humidity. If allowance is made for improvement of accident rates owing to increased precaution, as was done in examination of rainfall comparisons, corre- spondence becomes still more marked. The monthly comparison shows that the maximum concentration of accidents comes before the time of lowest humidity. Taken half-yearly, in accordance with seasons, the comparison is :— Accidents. Mean humidity. October to March....... 45 .... 67’3 April to September..... 60........ 65’4 If the six months of lowest humidity are put together, the comparison becomes :— Accidents. Mean humidity. January to June ....... 51 .... 71’1 July to December ...... 54 ....*. 61’6 It will be seen that the correspondence between the distribution of accidents in a year and the changes of humidity in that year is not very marked. We have now found (a) that both annually and seasonally the accident rates change in a reverse direction to the rainfall; (5) that seasonally they change in a reverse direction to the temperature; and (c) that annually they change in a reverse direction to the humidity of the atmosphere. The rainfall is directly connected with the interaction of temperature and humidity, and it remains to be seen whether by combining temperature and humidity in one considera- tion we get correspondence with accident rates both annually and seasonally. The only way in which temperature and humidity can affect colliery roofs is by drying or moistening. If the air entering the mine is cold and saturated, it absorbs moisture to keep it saturated as it rises in temperature. If it is hot and saturated it deposits moisture as it falls in temperature. If unsaturated when it enters the mine, it absorbs or deposits moisture according to the exact conditions with regard to moisture and temperature. The effect of air on a mine