The Interpretation of 
 Railway Statistics 
 
 Of 
 
 
 Ernest Ritson Dewsnup 
 
 Professor of Railway Administration in the Univer¬ 
 sity of Illinois 
 
 Reprinted from 
 
 Itoilroag^grfe'ttr 
 
 May 13, 1910 
 
^ o o . v\ Vj 
 
 D S'/S'* 
 
 [ttleprint from Railway Aye Gazette, May 13, 1910.] 
 
 THE NECESSITY OF CARE IN THE INTERPRETATION 
 OF RAILWAY STATISTICS USED COMPARATIVELY. 
 
 The growth in importance of the railway has given to its 
 statistics a publicity and a popularity hardly exceeded by those 
 of any other economic undertaking, and of no country is this 
 truer than the United States. Governmental activity has been 
 a stimulating cause of this abundance of information, but the 
 railways themselves have come to realize that it is good policy 
 to be less reserved in their relations with the public than they 
 have been. Thus, a much larger amount of information has 
 been made accessible than government schedules have called 
 for. The facts available, however, have not always been ana¬ 
 lyzed with due discrimination. Statistical material is highly 
 dangerous to handle, and even those familiar with its treat¬ 
 ment have to act with the utmost wariness and discretion. 
 Its way is not so plain that a wayfaring man, though a fool, 
 need not err therein. Defective comparison, arising out of 
 the omission of vital facts, is common. Certain more or less 
 traditional methods of presentation are followed that assume 
 to carry with them illuminating exposition of vital points of 
 railway development and operation, but are more remarkable 
 for what they fail to tell than for anything they do tell. 
 Brevity is the soul of wit in railway statistics as elsewhere, 
 but it can be pushed to the verge of obscurity and, quite 
 frequently, to the point of insufficiency. Another class of 
 defects in statistical comparison arises from a lack of appre¬ 
 ciation of the real significance of the data used: the statistician 
 fails to make himself acquainted with the intrinsic worth of 
 his figures. Nothing can be more inimical to the soundness 
 of. statistical logic than a hasty acceptance of material used 
 at its superficial value. It is absolutely essential to scrutinize 
 rigidly methods of compilation. Carelessness or oversight in 
 this connection is responsible for many erroneous analyses 
 and vicious comparisons. The following pages contain a 
 brief review of some of the more important railway statistics 
 in the use of which for comparative purposes the writer has 
 observed more or less serious errors. Some of the points 
 touched upon raise issues of considerable dispute, and, if it 
 seems that such are treated too cavalierly, it must be remem¬ 
 bered that this is inevitable in a general treatment of re¬ 
 stricted length; at a subsequent time it may be possible to 
 discuss these issues more in detail. 
 
 FREIGHT HOUSE TONNAGE COSTS. 
 
 A good example of the difficulties besetting the path of the 
 statistician without adequate technical knowledge of his facts 
 Is to be found in freight house tonnage costs. Freight house 
 operation seems to be very similar, in a general kind of way, 
 wherever it is carried on, and, therefore, fairly capable of being 
 measured as to its efficiency by the actual per ton costs of 
 
 3 
 
handling. As a matter of fact, however, while a ton of freight 
 is a uniform unit so far as mere weight goes, it is a very vari¬ 
 able one, indeed, with respect to its handling. One consignment 
 of freight may require twice as much handling as another of 
 equal weight. Thus a ton of general merchandise will consume 
 more of the working time of weighers, truckers and stowers than 
 a ton of iron bars. And a ton of merchandise in twenty differ¬ 
 ent packages will be a longer handling proposition than a ton 
 of similar freight in five packages. Freight houses called upon 
 to handle large quantities of mail-order house shipments 
 usually have this factor of increased cost to contend with, no 
 small amount of such business being in very small packages, 
 consigned to an innumerable variety of addresses. Dissimi¬ 
 larity in the nature of the freight handled manifests itself to 
 the highest degree in the railway terminals of districts with 
 different producing interests, but it may be found in terminals 
 located in the same district, and, to the extent that it exists, 
 renders tonnage costs entirely unreliable as a test of relative 
 efficiency. I have heard of this consideration being ignored 
 even by railway officers in comparing the tonnage costs of 
 freight stations at different points on the lines under their 
 jurisdiction. The officers concerned were perhaps relatives of 
 the newly-appointed division superintendent who, being re¬ 
 quested by shippers on his division to provide them with 
 double decker cars, objected on the ground of the expense that 
 would be involved in raising the bridges. 
 
 Another element with an important influence upon tonnage 
 cost is the size of the freight business handled. The small 
 freight house seems able to give an appreciable lead to its 
 bigger brother in handling costs, this arising largely from 
 the fact that the bigger freight house required for the busi¬ 
 ness of the latter increases the trucking that has to be done. 
 It ought to be possible, however, to offset much, if not all, of 
 this increase by the introduction of mechanical appliances in 
 the larger houses, but, so far, American freight houses are not 
 remarkable for their achievements in this direction. The 
 more important English terminals, it is interesting to note, 
 have made considerable progress in the utilization of hydraulic 
 and electrical power. 
 
 Where delivery of freight is undertaken or arranged for by 
 the railways, local conditions may cause, at some stations, an 
 abnormal amount of rehandling, which, of course, leaves its 
 mark upon the tonnage costs. Variations of cost, arising 
 from such cause, do not exist in this country to the extent 
 that they do, for instance, in England. In very rare instances 
 do railways here undertake the delivery of freight. But it is 
 essential to bear in mind this rehandling element of cost when 
 comparing American with English performances or the costs 
 of various English goods depots. At some important points 
 in England the amount of warehousing, as it is termed, is 
 considerably above the normal, an appreciable portion of the 
 
 4 
 
inward freight, instead of being sent out directly by the 
 wagons of the company or of its carting agents, having to be 
 raised by crane to a second, third, or even fourth, story, 
 trucked over a floor often of large area, with the reverse 
 operation to be gone through when the consignee calls for the 
 freight. Sometimes the agent is allowed extra tonnage on 
 this warehousing for the purpose of making costs comparable; 
 thus, for every ton warehoused, a ton and a half, say, would be 
 reckoned in computing the average cost, but this practice is 
 by no means universal. 
 
 Tonnage costs as between one terminal and another may 
 be affected by differences in the quality of labor available for 
 freight-house work. Should it happen that the labor environ¬ 
 ment of two terminals differed very materially, it would be 
 unfair to the local management at the less favorably situated 
 one to accept tonnage costs as indicative of relative efficiency 
 without adjustment being made for this difference. Occa¬ 
 sionally very special labor conditions introduce a still further 
 disturbing element. I cannot think of a better illustration 
 than the casual labor so effectively made use of at several 
 of the London goods depots. Where labor of this kind can be 
 procured but a minimum full-time or regular working force 
 needs to be maintained, the casual labor, waiting around the 
 office for a job, being brought in for the busy period at a 
 comparatively low rate of wage per hour. Ordinarily, such 
 labor would be regarded as unreliable and inefficient for the 
 class of work to be done, but in London this does not seem to 
 be the case. In fact, a considerable number of these men be¬ 
 come more or less regular “half-timers,” and vacancies in the 
 full-time force are filled from their ranks quite frequently. 
 Strange as it may appear it sometimes happens that such 
 promotion is refused, the chosen ones preferring the compara¬ 
 tively go-as-you-please conditions of casual employment to the 
 more rigid requirements of regular work. The result of such 
 labor arrangements is to enable the terminals thus advan¬ 
 tageously situated to keep their tonnage costs far below what 
 they would normally be. 
 
 Freight-house costs, it is hardly necessary to observe, may 
 be “in” or “out;” that is to say, the costs of working inbound 
 or outbound freight. The two are essentially distinct. Under 
 normal conditions, the cost of working inbound freight should 
 run lower than that of outbound, since the former freight can 
 be unloaded with greater dispatch and less irregularly than is 
 the case with the corresponding operations of loading. 
 
 Tonnage costs are commonly relied upon as a fair basis for 
 the comparison of efficiency in the same freight-house from 
 period to period, and, as a rule, justifiably so. But, even in 
 this case, the value of the comparison depends upon the 
 absence of change in the general nature of the tonnage han¬ 
 dled as well as on other conditions. But traffic conditions 
 alone being taken into account, it is quite possible that a 
 
 5 
 
change in these might increase tonnage costs, say, from 35c. 
 to 37c. a ton, in spite of a real improvement in the working 
 efficiency of the house during the period of comparison. 
 
 From what has been said it appears very evident that as a 
 comparative measure of relative efficiency of management the 
 tonnage cost of handling may prove to he a very misleading 
 guide, especially when applied to different terminals, but, at 
 times, even when applied to .the same terminal. 
 
 YAED CAE COSTS. 
 
 Yard car costs also require considerable care in their use. 
 As a test of relative skill of management, they fail because 
 of the differences in labor quality, in nature of work to be 
 done, in the physical condition of the yard, in climatic and 
 other conditions. But they may be, and frequently are, ren¬ 
 dered entirely useless for statistical purposes because of the 
 lack of a uniform system in compiling them. Thus, some ter¬ 
 minals base their yard costs per car upon the number of cars 
 received and forwarded in trains and received from and de¬ 
 livered to connecting lines. In other cases, not only these car 
 movements are included, but also many of the internal move¬ 
 ments, such as cars to and from transfer tracks, cars to and 
 from team tracks, and so on. This difficulty as to what prop¬ 
 erly constitutes a car movement could be overcome by agree¬ 
 ment, but at the present time it still presents a serious ob¬ 
 stacle to the successful comparison of yard costs as they are 
 actually furnished by different terminals. In every case, it is 
 necessary to go behind the figures given and to analyze their 
 method of compilation. 
 
 PASSENGEE AND TON-MILEAGE COSTS. 
 
 Passenger and ton-mileage costs have been made much use 
 of by practical men and by theorists, but, frequently, with 
 a careless disregard of the limits of their usefulness. More 
 than one writer has been content to assume the propriety 
 of measuring the general operating efficiency of one railway 
 system against that of another by means of the costs of opera¬ 
 tion per passenger or per ton per mile. Such a comparison 
 is too unreliable to be depended upon as a guide to relative 
 efficiency. It takes for granted that labor, physical, traffic 
 and other conditions are fairly similar, though, in fact, they 
 are frequently extremely dissimilar. To say that one road 
 operates its freight traffic at eight-tenths of a cent per ton- 
 mile and another road at six-tenths of a cent is to throw 
 absolutely no light upon the skill with which the freight 
 traffic of the respective roads is worked. The road with the 
 higher ton-mile cost may be the one most efficiently operated. 
 The only real value of such traffic mileage costs is in keep¬ 
 ing track of the movement of operating costs on the same 
 railway, or, preferably, on the individual divisions compris¬ 
 ing a railway. I say preferably, as regards the latter, be¬ 
 cause the general average for the railway as a whole may 
 have remained unchanged over the period of investigation 
 
 6 
 
and yet the real operating conditions may have materially 
 altered, progression on some divisions being counterbalanced 
 by retrogression on other divisions. Even within the limits 
 of a single railway it may be decidedly unfair to measure 
 the operating results of one division against those of an¬ 
 other—conditions must be substantially similar before cost 
 results can be accepted as the basis of such a comparison. 
 
 The value of ton-mile costs has been emphasized by some 
 because of a belief that they afford a scientific basis for the 
 establishment of reasonable rates. Yet, it is not possible to 
 allocate, other than by arbitrary pro-rates, more than, at 
 the most, about 75 per cent, of even the transportation and 
 maintenance costs to the respective freight and passenger 
 services. And since, as a matter of practical necessity, a 
 railway must pay interest on funded debt, interest on current 
 liabilities, rents for leased lines, taxes, and, for future pro¬ 
 tection, permanent improvement expenditures charged to in¬ 
 come, there still remains nearly 30 per cent, of the total 
 expenditures, esrential to the working of the system, which 
 again can be allocated only on a more or less arbitrary pro¬ 
 rating basis. Thus, close upon half of the annual costs of 
 maintaining a railway organization and securing its opera¬ 
 tion are, in the present state of our knowledge of railway 
 costs, incapable of being accurately divided. The actual cost 
 might vary materially from the pro-rated cost, and, to the ex¬ 
 tent of this variation, the reliability of the latter as a crite¬ 
 rion w r ould be impaired. 
 
 But even if it were practicable to allocate to the freight 
 service as a».whole its proper proportion of costs, it would 
 still be a far cry to the distribution of these costs between 
 the multitude of commodities transported by the railway 
 under very varying conditions of speed, equipment facilities, 
 terminal handling and so forth. There are some happily-dis- 
 positioned writers and regulative bodies who see all these 
 difficulties vanishing into thin air when resort is had to some 
 magical pro-rating basis as train mileage or gross ton-mile¬ 
 age, or the like. These engineers of philosophy are so much 
 in advance of many of us poor mortals that they have no 
 trouble in determining the roadway maintenance expense due 
 to a 50-mile an hour 500-ton through express passenger train? 
 as compared with that due to a 12-mile an hour 2,500-ton 
 freight train, stopping, maybe, at almost every siding. Their- 
 attitude is picturesque, but unconvincing. 
 
 It should be said further that if an allocation of costs to; 
 each unit of freight could be satisfactorily made, the problem 
 of reasonableness would still remain unsettled. The theory- 
 of reasonable rate determination with reference to traffic mile¬ 
 age costs means, I take it, the fixing of a rate at a figure al¬ 
 lowing a reasonable rate of profit above cost of operation; 
 but, as a practical proposition, how is the latter to be de¬ 
 termined? It should certainly be reasonably related to the 
 
 7 
 
risks and difficulties of the business, but who is to fix the 
 specific nature of that relation? Are the risks of railway 
 business equal to, less than, or greater than, those of bank¬ 
 ing, insurance, steel manufacture? If not identical, what per¬ 
 centage of variation in profits will represent the variation in 
 risk? Again', absolutely uniform efficiency of management 
 cannot be expected, and uniformity of working environment 
 is, of course, impossible. Hence, one series of railway opera¬ 
 tions under control of railway company X will be handled 
 much more economically than a corresponding series con¬ 
 trolled by railway company Y, resulting in a lower cost per 
 unit of traffic. Is X to be allowed to charge rates similar 
 to Y, and, if so, is the extra large profit it secures as a re¬ 
 sult of its greater efficiency a reasonable rate of profit? On 
 the other hand, if its rates are adjusted so as to reduce its 
 rate of profit to the same level as that of Y, will there not 
 result territorial rate discrimination quite as clear-cut as any¬ 
 thing that occurs under the present commercial system of 
 rate making? 
 
 TRAIN AND ENGINE-MILE AVERAGES. 
 
 Other interesting operating statistics are train and engine- 
 mile averages, which, in their general form, have less utility 
 than some would ascribe to them. A passenger train-mile 
 covers everything from the more or less dilapidate^ crawling 
 local on branch lines to 60 miles an hour luxuriously ap¬ 
 pointed expresses on the main line, from a train of an engine 
 and one car to one with a dozen or more cars. Similarly, the 
 freight train mile coerces into statistical uniformity the 
 humblest way freight with, perhaps, a total weight of a couple 
 of hundred tons behind the tender to the lOlT-car, 3,000-ton 
 train; it connects the slow “dead” freight train with the high¬ 
 speed “manifest” train: coal train, live stock, merchandise 
 train—all are indiscriminately mingled in the average. An 
 arithmetical average is obtained, it is true, but one of treach¬ 
 erous utility, unless environed by a galaxy of interpreting, 
 modifying and saving clauses, only to be used by the initiated, 
 and even then with the greatest circumspection. 
 
 It is still customary, on the part of some transportation 
 students, to depend upon engine mileage as a reliable basis 
 upon which to establish comparative efficiency statistics of an 
 important brand of railway operation. The value of such 
 statistics depends very largely upon the way in which they 
 are put together. But general averages of engine-mile costs 
 can hardly be regarded, under any conditions, as affording 
 the means of drawing conclusions of any practical worth. 
 The engine-mile, in its general form, is itself variable. Its 
 elements may be freight or passenger or switching engine, 
 high-speed Atlantic or low-speed consolidations, with tractive 
 powers of infinite variety. Obviously, when such engine mile 
 age is used as a measure of costs, it fails lamentably to serve 
 any useful purpose. To do this it must be limited to engines 
 
of the same general class, and even then its definiteness is 
 not altogether satisfactory. Clearly, a compound unit of per¬ 
 formance which allows for the influences of physical con¬ 
 ditions, such as gradient and curvature, and also for speed, is 
 the only sort of unit really suitable even in a restricted 
 analysis of engine mileage costs. Uniform units of work done 
 need to be established and engine costs related to these; this 
 the engine-mile average fails to do. 
 
 FREIGHT CAR TON MILEAGE. 
 
 Freight car efficiency is a matter of great railway interest 
 and its proper measurement seems to be logically based upon 
 the performance of the average car in loaded car miles. The 
 calculation of this would ordinarily be a very simple process, 
 namely, the division of the total ton mileage by the number 
 of cars employed. Unfortunately for the student of railway 
 reports, it is rot obtainable so easily as one would suppose 
 from the foregoing. In the first place, the freight cars re¬ 
 ported as ‘owned, both in the statistics furnished in the annual 
 reports of the Interstate Commerce Commission and in very 
 many railway reports, are those owned on the last day of the 
 fiscal year. It is obvious that the loaded car miles are made 
 by the actual cars in operation, some of which have been in 
 service throughout the year, others for less periods varying 
 from several months to but a few days. Evidently, the equiva¬ 
 lent number of cars owned throughout the year is required 
 as the divisor, otherwise the quotient will be more or less 
 inaccurate as a test of efficiency. In the second place, the 
 ton mileage reported is the result of work done both by home 
 and foreign cars. Cases have been known in which the cars 
 upon the lines of a railway have exceeded b^ as much as one 
 hundred per cent, the number owned by it. Under such con¬ 
 ditions, even if the equivalent number of cars owned through¬ 
 out the year were available, the quotient of total ton mileage 
 divided by the same would be very far from representing the 
 exact work of the cars. Of course, in dealing with statistics 
 of the United States as a whole, this possible source of error 
 disappears except in so far as car relations with Canada and 
 Mexico affect the situation, and any disturbance of results 
 from such would be so small as to be negligible. In the third 
 place, reported ton mileage is made up of tonnage hauled by 
 the railway both in its own cars and in privately owned cars. 
 Here again it is apparent that the car ton mileage obtained 
 in the usual fashion is in error. The private cars working 
 the ton mileage need to be included, or better, for purposes 
 of the comparison of efficiency of the freight car operation of 
 different roads, the ton mileage of private cars excluded. The 
 latter is preferable because of the special conditions surround¬ 
 ing certain classes of private car movement, ensuring fre¬ 
 quently a superior performance to that of the railway owned 
 car, and, further, because of the differences, as between rail¬ 
 ways, in the extent of this private car traffic. 
 
But supposing that all these difficulties in the determination 
 of freight car performance are satisfactorily met, the greatest 
 caution is still required in interpreting the results. It is quite 
 conceivable that a lower ton mileage average may represent 
 as great or greater operating efficiency on the part of the rail¬ 
 way concerned as the higher one of another road. This arises 
 out of the varying influence on mileage of different species of 
 traffic. The larger the proportion of local freight traffic, the 
 lower will tend to be the mileage made, but the reduced ton 
 mileage average by no means infers less efficiency. Again, the 
 existence of an unusually large percentage of colliery traffic 
 will ordinarily exercise a depressive influence upon the ton 
 mileage, on account of mine delays and unloading time allow¬ 
 ances. Demurrage regulations are also a factor not to be 
 overlooked in deducing results from ton mileage performances, 
 of railways situated in different territories. If the legal loan¬ 
 ing or unloading free time is 48 hours in the territory of one 
 road and 96 hours in the territory of another, mileage move¬ 
 ment in the latter case is restricted. The restriction is not 
 effective to the extent of the difference in legal time, it may 
 be noted, since there would be in both territories a large num¬ 
 ber of shippers who would not make use of the legal maximum. 
 
 THE OPERATING RATIO. 
 
 The operating ratio is generally accepted as a statistic of 
 great value. Yet in the usual form of the percentage of total- 
 annual operating expenditure to total annual operating rev¬ 
 enue, its value is liable to exaggeration. The same operating 
 ratio may conceal any number of changes in efficiency of per¬ 
 formance from year to year. Thus, in a certain year, extrava¬ 
 gant operation during the season of light traffic may be coun¬ 
 terbalanced by an abnormal increase of business during the 
 period of heavy traffic, with the result that the operating ratio 
 stands at the same level as, or even lower than, in the previous 
 year when, say, greater uniform efficiency of operation was 
 obtained throughout the year. The general operating ratio 
 may frequently fail to indicate movements in maintenance 
 of roadway, maintenance of equipment, traffic, conducting 
 transportation, and general administration, in which consid¬ 
 erable variations may occur and yet leave no record in the 
 general ratio. The mere statement that the operating ratio 
 is this and that in years of comparison is meaningless from 
 the point of view of operating efficiency until most careful 
 scrutiny is made of its component factors. An increase in it 
 may cover greater efficiency, a decrease relatively greater in¬ 
 efficiency, and a stationary position either one or the other. 
 
 Error can easily be made in comparing operating ratios of 
 different systems by reason of opposed financial policies in the 
 roads compared. For example, the ratios are not comparable 
 in the case of two systems, one of which charges as much of 
 permanent improvement expenditure as possible to repairs 
 account and the other as little as possible. 
 
 10 
 
PASSENGER ACCIDENT AVERAGES. 
 
 Passenger train accident figures are of all railway statistics 
 the ones which are most profusely presented to and .most 
 eagerly devoured by the public. It goes without saying that 
 the presentation of numbers of killed and injured without 
 reference either to the conditions out of which the accidents 
 arose or to the relation of these numbers to the traffic is a 
 very partial and defective, therefore improper, method of 
 comparing safety in travel on different systems. Nor are 
 such averages as number killed and number injured per so 
 many passenger journeys of appreciable value in endeavoring 
 to measure the relative frequency of accident. 
 
 It is necessary, in the first *place, to segregate accidents 
 arising out of passenger train movement and beyond the con¬ 
 trol of the passenger, from the rest, and, in the second place,, 
 to average these accidents upon the passenger train mileage, 
 so many accidents per one million train miles, and, associated 
 with this average, should be the average number, killed and 
 average number injured pe‘r one million train miles. The 
 latter figures emphasize the waste of human life and efficiency, 
 but to my mind the former averages are the more valuable. 
 The number killed and injured in a collision or derailment is 
 almost a pure matter of chance, influenced by car structure, 
 by presence of inflammmable oils, live coals, etc., by distribu¬ 
 tion of passengers in the cars, and by other circumstances. 
 An accident that, by all theory, ought to have caused the most 
 serious disaster may be trifling in its effect; an accident that 
 normally would have produced a few injuries may, unex¬ 
 pectedly, result in lamentable loss of life. So that really, 
 relative efficiency, in the matter of avoiding accidents in pas¬ 
 senger traffic, would be shown best, not by reference to persons 
 killed and injured, but by the number of train accidents per 
 million passenger train miles, and such figures would form a 
 desirable addition to every report of railway accidents. Rail¬ 
 ways have a natural aversion to display their weaknesses in 
 the form of train accidents, but it would exert a beneficial 
 influence upon them if they were required half year by half 
 year to publish in the leading daily papers a full statement 
 on oath showing the relationship of their accident roll during 
 the period to the amount of traffic they handle. At the same 
 time it would be unfair to the railways to pass on without 
 saying that several roads have been and are devoting great 
 attention to the removal of the causes of accidents. Surprise 
 checking, public investigations, and so on, are some of the 
 methods that promise satisfactorily. 
 
 MILEAGE STATISTICS. 
 
 Little need be said here about mileage statistics. It is obvi¬ 
 ously inaccurate to compare the extent of railway systems by 
 route, or geographical, miles. Such a comparison ignores the 
 capacities of the systems compared, as represented by addi- 
 
 11 
 
tional tracks and varying yard facilities. It is like comparing 
 two factories on the basis of the number of machines em¬ 
 ployed, without reference to the capacity of the same. Of late 
 years this has been rectified in some private and official statis¬ 
 tics by showing, in addition to route mileage, single track 
 mileage, and, in some cases, mileage of yards and sidings. 
 Even when this information is given, further analysis is re¬ 
 quired if jone is to gain a useful idea as to the real meaning 
 of the mileage to the community. When comparing the mile¬ 
 age of railways or railway systems, the student needs always 
 to ask himself the question as to the extent to which effective 
 distribution of mileage has been attained, and effectiveness in 
 this case should refer to tne general economic interests of the 
 country, for, after all, sucn interests in the long run underlie 
 private progress and success. The distribution of the mileage 
 is an important consideration, since, in two regions of equal 
 area and equal railway mileage, the mileage may be well 
 spread in one and very badly spread in the other. It is to 
 the ultimate economic interests of any state or country that 
 its railway mileage should be located so as to give reasonable, 
 though not necessarily uniformly equal, accessibility to all 
 districts. Over-concentration of mileage in one part, a not un¬ 
 common phenomenon, is frequently accompanied by lack of 
 facilities in other parts. Again, an area intermediate between 
 two other areas may have a fairly considerable railway mile¬ 
 age much more calculated to serve the interests of the other 
 areas than those of its own. Mileage per unit of area or per 
 unit of population is a useful figure, hut incomplete. Equal¬ 
 ity in these respects may hide very real differences of utility 
 and an area with an inferior showing may really be better 
 served. 
 
 TRAFFIC MEASUREMENT. 
 
 limits of space forbid much attention in this article to the 
 statistics of traffic measurement, though the subject is one 
 worthy of detailed discussion. It is generally understood that, 
 for purposes of comparison of freight business, ton-mileage is 
 a more adequate basis than actual tonnage. Yet even ton-mile¬ 
 age is unsatisfactory; its general nature necessarily deprives 
 it of efficiency as a measure where conditions are different, as 
 is commonly the case. The tonnage movement measured may 
 be largely one of minerals “foreign received” and “foreign 
 delivered,” or it may he largely one of merchandise locally 
 collected and distributed. In amount of work involved, in 
 value of service to the railway, the traffic of one road would 
 he poles apart from that of the other, but no indication what¬ 
 ever of these differences would find expression in the general 
 ton-mileage figures. In every case of summarized statistical 
 presentation of freight traffic conditions it should be made an 
 invariable rule to show grouped commodity ton-mileage; in 
 most instances the requirements of general comparison would 
 be met by a classification into forest products, mine products, 
 
 12 
 
agricultural products, manufactures, merchandise and miscel¬ 
 laneous. From statistics of this kind one is able to form a 
 fairly intelligible idea of the traffic of roads compared and 
 their relative importance in this respect, which is more than 
 can be said of the lumped ton-mileage figures. A further divi¬ 
 sion might be made into local, through-originating, through- 
 terminal and through-transfer, understanding by these terms 
 the ton-mileage of freight forwarded from and delivered to 
 stations on company’s system (local), forwarded from local 
 stations to “foreign” points (through-originating), received 
 at local stations from “foreign” points (through-termnal), and 
 received from “foreign” stations for delivery to “foreign” 
 stations (through-transfer). This would complicate some¬ 
 what any general tabulation of statistics, but not seriously, 
 when the information deducible from the addition is taken 
 into account. Briefness of exposition and economy of space 
 are idle virtues if they are secured at the expense of intelligi¬ 
 bility. 
 
 CAPITALIZATION STATISTICS. 
 
 Capitalization per mile is a statistic which incompetent 
 hands are capable of badly *mis-using. Intrinsically the figure 
 is worthless, for comparison useless, unless qualified by a 
 whole host of explanatory clauses. Yet it is frequently used 
 in comparing costs of different systems. The practice is 
 faulty because (1) the comparison is made on geographical 
 mileage whereas the expenditure has been made on track 
 mileage, (2) the manipulation of capital in the form of nomi¬ 
 nal additions has varied considerably from system to system 
 so that the ratios of real costs are very different from those 
 of nominal costs, (3) the existing extent of capital has been 
 influenced to no small degree by differences in policy as to 
 the handling of betterment expenditures. In any case, since 
 conditions of legal procedure, land acquisition, labor supply 
 and so forth, vary most widely in railway building, it is hard 
 to see how much wiser one is for knowing that Railway X 
 cost $70,000 a mile and Railway Y $250,000, even when ad¬ 
 justment has been made for artificial differences in capitaliza¬ 
 tion. Certainly, without a very detailed knowledge of the 
 underlying conditions, it would be hazardous to assert that 
 the former road was economically built and the latter ex¬ 
 travagantly. The costliness of a machine is relative to the 
 work it does, and hence, of the two, the $70,000 road may 
 happen to be the extravagantly built one. 
 
 Capitalization statistics evidence great variations in capital 
 per mile of line, and this fact, correlated, as it has been, with 
 spectacular feats of financial legerdemain on the part of cer¬ 
 tain railways, has wrought a conviction in the minds of many 
 that overcapitalization is a common condition of American 
 railways, influencing unfavorably the level of rates. Now it 
 Is unquestionable that considerable nominal additions have 
 been made to railway capital, but against such increases need 
 
 13 
 
be set off the also considerable portions of revenue that have 
 been expended in permanent improvement, as well as those 
 decreases in capital that have occasionally occurred in trans¬ 
 fer of ownership. It is impossible to determine the relative 
 size of these items, and he is a man of easy conviction who 
 is prepared to argue that, for the railways of the United 
 States as a whole, the former is materially greater than the 
 latter. And if it is, the difference must be very much less 
 than most people credit. The previous statement infers that 
 the proper measure of railway capitalization is actual orig¬ 
 inal investment in the property plus the cost of improvements 
 whether paid for out of capital or out of revenue. If rail¬ 
 way building were to be started all over again, the restric¬ 
 tion of capitalization within these limits would be justifiable, 
 theoretically, at any rate. It is to the interest of the in¬ 
 vesting public, and of the community at large, that the real 
 profit capacity of every railway property should be apparent 
 on the face of its balance sheet and income account; only 
 when capitalization is identical with actual investment, is 
 it practicable to keep track of the real profitability of the 
 undertaking. It would surely seem that actual investment in 
 the building and improvement of a railway is the logical 
 basis of capitalization, though it may be a very untenable 
 one from the standpoint of business men and their traditional 
 methods. But, comes the objection, is it not proper, even 
 theoretically, to allow X or Y to increase the capitalization 
 of a railway he purchases up to the limit of the price actually 
 paid? Does not this price represent to him a real invest¬ 
 ment? Plausible though this seems, I can hardly admit that 
 it is theoretically justifiable. What X or Y pays to secure 
 the profits to which each certificate of the capitalization en¬ 
 titles him represents his valuation of the property; thereby 
 he decides with what rate of profit he will be satisfied. The 
 money he pays goes, not into the railway, but into the pocket 
 of its former owners. There is no valid reason why his valua¬ 
 tion and what he purchases, namely, the investment of a 
 million dollars, or whatever it may be, in the production of 
 a certain utility, should be equalized. In fact, logically they 
 cannot be. The investment for the purpose of producing the 
 utility, and it is immaterial whether some part of this in¬ 
 vestment was judiciously made or not, is one thing, and his 
 investment for the purpose of securing the profits arising 
 out of that utility is another. As a matter of fact, the pur¬ 
 poses underlying increase of capitalization, following acquisi¬ 
 tion of a property, are frequently those relating to speculative 
 trading or to the concealment of rate of profit. This does 
 not deny the justification, either at the time of a purchase, 
 or at any other time, of an increase in capitalization equal to 
 those permanent improvement expenditures out of the revenue 
 of the property not previously taken into capital account, or, 
 of course, of an increase in capitalization for the purpose of 
 
 14 
 
securing funds which are intended to be, and actually are, 
 sunk in the business. I fully realize that the application of 
 the theory flow stated would have retarded the promotion 
 and construction of certain railways, but the history of rail¬ 
 way enterprise in this country, especially for the couple of 
 decades or so following the Civil War, leads one to believe 
 that this retardation might have had advantages of its own, 
 and that there would have been less economic loss from fool¬ 
 ish location and unduly speculative building. 
 
 A business undertaking is sometimes capitalized at an 
 amount over and above that necessary to carry on the busi¬ 
 ness,-only a part of the amount called for by each share being 
 actually subscribed, this arrangement having advantages in 
 enabling further capital (within the limit of the share lia¬ 
 bility) to be obtained with certainty, promptness, and fre¬ 
 quently with economy. In such cases, the real capital is the 
 subscribed capital, and the law should require every official 
 statement of capitalization of such a company, and every stock 
 exchange quotation, to be accompanied by figures showing the 
 subscribed value of each stock certificate. 
 
 Sufficient has been said to indicate the rashness of a sweep¬ 
 ing charge of overcapitalization, since it is clear that the 
 only real and satisfactory test of capitalization cannot be 
 applied. This difficulty has induced financial writers to sug¬ 
 gest the application of cost of reproduction or of market 
 valuation of securities as suitable criteria. The latter is but 
 a function of income-earning power, representing an averaging 
 down of profits, and bears no relation to the investment. Capi¬ 
 talization on this basis serves but to hide the real profitability 
 of the investment. And, anyhow, the market prices of securi¬ 
 ties bought and sold on the exchange form an unreliable guide, 
 because these are the prices usually of but limited amounts, 
 and it by no means follows that an offering of the whole issue 
 at one time would secure similar prices. Cost of reduplica¬ 
 tion is obtaining more favor than market valuation, but, ob¬ 
 viously, as costs and conditions of construction are constantly 
 changing, the cost of reduplication may be far removed from 
 the actual costs, and hence this fails as. an accurate criterion. 
 Yet in the case of the majority of American railways, a phy¬ 
 sical valuation of the existing property would probably rep¬ 
 resent a reasonably fair maximum capitalization, and capi¬ 
 talization much in excess of this, though not necessarily ficti¬ 
 tious, must be looked upon with suspicion. If phy:icai valua¬ 
 tion is to be accepted as a satisfactory test of capitalization, 
 the charge of general overcapitalization must forthwith disap¬ 
 pear, as it may safely be said to be impossible to reproduce 
 the railways of the United States for the amount per mile rep¬ 
 resented by their capitalization. 
 
 The effect of overcapitalization upon rates falls too much 
 outside the scope of this article to justify more than a brief 
 reference. It is difficult to imagine that, under the competi- 
 
 15 
 
tive conditions so characteristic of a large part of the rail¬ 
 way business of the United States, degree of capitalization 
 can definitely influence transportation rates. The capital in¬ 
 vested in railways is largely sunk once and for all. Track, 
 locomotives, cars and appliances will deteriorate, even if not 
 used. Hence, any rate that returns, in the estimation of the 
 railway manager, somewhat more than sufficient to defray 
 wear and tear due to the traffic and cost of transportation 
 will be better than no traffic. This is, under ordinary condi¬ 
 tions, the nether limit of rates, a limit to which they rarely 
 descend, but the existence of which, neverthelezs, is powerful 
 iu influencing the actual level at which they settle. Such fav¬ 
 orable conditions for the exercise of competition prevent the 
 railway manager from basing his rates upon the capitaliza¬ 
 tion. Thus it happens that, in the case of a group of im¬ 
 portant western roads, the one with more than three times 
 the capitalization of any of its rivals is forced, by the exi¬ 
 gencies of competition, to give even lower rates than the 
 others. Where a condition of monopoly exists, it is possible 
 that overcapitalization may have a general effect upon rates, 
 if the rates, before the excess of capital is superimposed, are 
 such that an appreciable increase in them would neither 
 lessen present traffic nor prevent the development of future 
 traffic to any appreciable extent. If the railway, under such 
 monopolistic conditions, is already exacting the return which 
 gives it the maximum net profit, then an increase in rates 
 would be inconceivable, as it would slow down the traffic and 
 represent a voluntary reduction of profit to a point below the 
 maximum. The conditions under which cases of capitaliza¬ 
 tion tend to affect the general level of rates are much more 
 likely to occur in local than in through traffic, but even local 
 rates fail to show many instances of increase that can be 
 definitely ascribed to manipulation of capitalization. 
 
 The foregoing survey of some of the more important rail¬ 
 way statistics has been by no means exhaustive, but enough 
 has been said, it is hoped, to emphasize the Jimitafion attach¬ 
 ing to them, especially as usually presented, and the necessity 
 of a careful and intelligent interpretation. .Neither railway 
 officials nor writers employing such statistics have always 
 exercised good judgment in this respect. Too frequently may 
 they be found resorting to reasoning which suggests, as a 
 well-known statistician remarked some years ago, “a man 
 trying to lift himself over a fence by his boot straps.” It 
 is, indeed, extremely easy in statistical elaboration to pass 
 from one assumption to another until out of a mass of figures 
 certain results are obtained which are accepted as precise and 
 accurate without recognition of their hypothetical basis. Only 
 by extreme vigilance can errors of this kind be avoided. In 
 so far as the previous discussion may act as a reminder of 
 this “proneness to error” in statistical reasoning, it will have 
 served its purpose. 
 
 16