Jason Bourne

At its beginning the United States was so deficient in avenues of transportation—with roads in some areas practically impassable several months of the year—that political disintegration was gravely feared. So insistent were the demands for improvement that, acting on a Senate resolution of 1807, Secretary of the Treasury Albert Gallatin prepared an analysis and program, presented in 1808.

He urged the national government to build a series of canals along the Atlantic seaboard from Massachusetts to the Carolinas; build interior canals and roads; and establish communication between the Atlantic and midwestern rivers and with the St. Lawrence Seaway and the Great Lakes. He thought that all of the improvements could be made for $20 million, and as the Treasury was steadily accumulating a surplus, that the debt could be paid in ten years.

This proposed indebtedness, the first suggestion of the sort in U.S. history, was bitterly denounced by many, and President Thomas Jefferson did not believe the idea constitutional. While the subject was being debated, the War of 1812 approached and soon stopped all thought of the projects. After the war they were brought up again, and four roads were built, but no canals.

Gallatin’s report was prophetic in that most of the works he advocated were later completed either by the federal government, as was the Intracoastal Waterway, or by the states, as was the Erie Canal. The subject of internal improvements became increasingly divisive during the antebellum period, pitting Whigs, who generally supported federal funds for transportation improvements, against Democrats, who did not. Bibliography

Ewing, Frank E. America’s Forgotten Statesman: Albert Gallatin. New York: Vantage Press, 1959. Kuppenheimer, L. B. Albert Gallatin’s Vision of Democratic Stability: An Interpretive Profile. Westport, Conn.: Praeger, 1996.

Walters, Ray. Albert Gallatin: Jeffersonian Financier and Diplomat. New York: Macmillan, 1957.Cost and Safety Efficient Design Study of Rural Roadsin Developing CountriesFinal Reportby B L Hills, C J Baguley and S J KirkEXECUTIVE SUMMARYThe DFID/TRL guide “Towards Safer Roads in Developing Countries” (TRL,1991) was produced by road safety engineers to help change the attitudes of highway planners and engineers in developing countries to the needs of road safety – in particular the needs of the pedestrian. It has been very successful in raising the awareness of the problem around the world, but there remains a reluctance amongst many planners and engineers operating in developing countries to adopt the recommendations.

The reasons for this are a combination of (i) a lack of detailed design information of alternative safety features; and (ii) a common belief that incorporating safety is expensive, and that with limited funds available, this will result in a shorter length of road being built. This project was designed to bridge the gap between the needs of cost conscious and safety conscious highway planning and design in developing countries. It has attempted to identify those aspects of highway planning and design where safety can be maximised at minimal cost, with the aim of producing practical Cost and Safety Efficient design notes (CaSE notes) to show how this can be achieved. It was planned that these would eventually lead to the running of in-country training seminars in a later phase.

Project Methodology1. Existing data – A safety highway design study re-visited: As part of a programme of collaborative road safety research between the TRL (funded by DFID) and the Papua New Guinea Department of Transport, a geometric survey was carried out in 1990 of the Highlands (Okuk) Highway from Mt Hagen to Lae, a distance of 480 km. It particularly examined features such as drainage ditch design and side slope profiles that could possibly be a factor in the very high levels of rollover accidents. Because of the detailed nature of the geometric design survey and the good accident data, it was considered appropriate for this present study to re-examine the data using Generalised Linear Modelling (GLIM).

2. New data – Video, Accident and Cost studies in six countries: Because there are major differences between the design and operational conditions of rural roads in Sub-Saharan Africa and the Indian Sub-continent, the project focussed on both regions. The countries selected were based on a combination availability of MAAP(Microcomputer Accident Analysis Package) accident data for rural highways, a wide range of highway designs and operating conditions, and the use of innovative or marginal designs.

The countries selected were therefore Zimbabwe, Botswana, Tanzania, Malawi, India and Nepal. There were three areas of data collection: (i) highway surveys; (ii) accident data collection; and (iii) highway design drawings and detailed breakdowns of construction costs. To maximise the efficiency of any survey work undertaken, it was decided to make extensive use of a digital video logging system. 3. Special studies

a) Three roads in Tanzania with widely differing geometric standards.This examined the safety records of three trunk roads meeting at Chalinze whose standards vary from 1.5m shoulders with 1:3 sideslopes to 0.5m shoulders with 1:1.5 sideslopes. b) Accident countermeasures in Papua New Guinea.Three projects carried out on the Highlands Highway are analysed here for the first time: over 10km of footpath constructed alongside the Highway on the approaches to two towns; chevrons boards erected on bends causing a high level of rollover accidents; and yellow bar pattern painted on the approaches to single lane bridges. Summary of Findings

1. Papua New Guinea Highlands Highway studya) The most significant terms in the models predicting accidents were always found to be Average Annual Daily Traffic (AADT) and Section Length. b) A non-linear relationship was found for Overturn accident rate v Average Shoulder Width whereby accident rates fall with increasing Average Shoulder Widths up to about 1.5m, and then accident rates increase with further increases in shoulder width.

Widened shoulders were often found on just one side of the road associated with ribbon development of some sort; an absence of any marking, kerbing or other physical definition of the widened shoulder could be noted at a number of these sites. An analysis of written police reports revealed two types of overturn accident: (i) ‘Too Fast’ or ‘Loss of Control’; and (ii) ‘Avoiding Other Vehicle’. Whilst the first type appeared all along the Highway, the second type only appeared near urban areas in the sections where the wide shoulders occurred.

c) There is generally more scatter for the other collision types, but there is some evidence that similar non-linear relationships for Head On and Side Swipe accidents exist as for Overturn accidents d) For the collision types, Overturn, Head On and Side Swipe, for the Narrow Shoulder width group of data (<1.5m), a single 90degree bend in a road section was found to be the strongest highway geometric term predicting accidents. e) It had been speculated that Overturn and Hit Object Off Road accidents were very similar in cause and would have similar models. The analyses suggested that this is not the case, with the distribution of the two types of accidents showing systematic differences.

f) For ‘Hit object off road’ accidents, Gradient is the strongest geometric term in the model, with multiple 90degree bends in a road section being the only other geometric term to appear in the model. g) For Pedestrian accidents, the ‘snapshot’ Pedestrian Count, taken as the survey team moved from one survey point to the next, was found to be strongly significant. The geometric term Average Drainage Ditch Depth and the interaction between Average Shoulder Width and Gradient (positive in direction) were also found to be significant terms. Highway curvature of any type has no effect on Pedestrian accident rates on the Highlands Highway. 2. Three countries study – Zimbabwe, Nepal and India

a) The GLIM analyses showed it was necessary to develop separate models for Zimbabwe on the one hand and India and Nepal on the other hand, reflecting the very different nature of the traffic between the two continents. Malawi and Tanzania were not included in the analyses because of difficulties in obtaining good data, and Botswana eventually had to be dropped due to uncertainties in accident locations. b) Generally these models did not produce as good a fit as those found for the Papua New Guinea data, ie. the percentages of the variance explained by the models were noticeably lower.

This may be partly be due to the greater number of years of accident data and a greater proportion of more objective factor-type data being taken from video than actual measurements as was the case in Papua New Guinea. c) Again the most important explanatory variable for All accidents was traffic flow, and both the African and Asian sample roads exhibited almost direct proportionality; that is, the power of the traffic flow term is almost unity. d) Increasing carriageway width was found to reduce accident rate, although increasing shoulder width was, surprisingly, found to increase accidents (as was found in the Papua New Guinea model for shoulders in excess of 1.5m).

However, carriageway width had the most influence such that if both are increased by, say, 0.5m, the net effect would be a predicted reduction in accident rate. Also, it is likely that wider shoulders are normally associated with different land use (ie. not simply vegetation) and thus increased pedestrian and other road user activity, which in itself tends to lead to higher accident rates.

e) In Zimbabwe, in most accident type models, the presence of a bus bay seems to lead to an increase in accident rate, despite the bays generally being well designed. It is recommended that a behavioural study be carried out to try to determine factors that may be contributing to accident occurrence near bus bays. f) For several accident types in Zimbabwe, such as Rollover, Hit Object Off-road and Rearend, the presence of a solid centre line was found to be associated with an increase in accident rate. For the All accidents model this was also found to be true if a Vehicle Restraint has been installed at the edge of the carriageway.

However, these features are not thought to be increasing danger but are simply indicative of the road situation as they are normally installed on hazardous bends or elsewhere where sightline is reduced or there is a severe drop from the carriageway. In the Nepal/India model where Curvature was included from the geometric drawings, there was clearer indication that the sharpness of a bend generally increases accident rate as this is a significant variable in the model. g) The Pedestrian accident rate in Zimbabwe is strongly related to the level of pedestrian use of the road (‘snapshot’ Pedestrian Count). Pedestrian accident rate also appears to have a similar relationship to Carriageway Width and Shoulder Width as for All accidents stated in (d) above.

h) For the Nepal/India dataset it was found that a reasonable model fit could be made for All accident types but that the numbers of individual accident types were too small to produce reliable individual models. Curvature and Gradient proved to be significant explanatory variables, both increasing accident rates the sharper the curve or steeper the gradient. For Papua New Guinea study, it was the interaction between these variables that was highly significant for Overturn accidents. i) According to the model, the presence of a marked Edgeline in Nepal and India appears to be particularly beneficial in reducing accident rate. 3. Special Studies

a) Whilst no evidence could be found that wider shoulder widths, as constructed on the PNG Highlands Highway, had any benefit for pedestrians, 10kms of segregated footpaths constructed alongside the Highway were found to be highly cost-effective, with estimated First Year Rates of Return in the range 400-1000%. Also, in comparing pedestrian accidents on two rural highways in Tanzania, it was found that the pedestriancasualty rate on the road with only 0.5m shoulders was 116% higher than on the road with 1.5m shoulders when the pedestrian counts/km were only 36% higher.

b) In comparing non-pedestrian accidents on the two rural highways in Tanzania, it was found that the casualty rate on the road with only 0.5m shoulders and 1:1.5 sideslopes was 52% higher than that of the road with 1.5m shoulders and1:3 sideslopes. c) The installation of chevron boards on bends on a section of road experiencing a high level of single vehicle night-time accidents were found to be highly cost-effective with accidents reduced overall by 44 per cent. d) It was found that the painting of yellow bar patterns on the approaches to single lane bridges on the Highlands Highway in Papua New Guinea reduced accidents by at least 25%.

The approaches frequently involved a sharp bend. e) On the Ramu Highway in Papua New Guinea, the introduction of the yellow bar patterns and chevron boards at 90 degree bends on the approaches to two bridges reduced a dramatic sharp rise in accidents (six-fold increase in one case) following the sealing of the road, back down to, approximately, the original levels. f) The sealing of 15km of unsealed road on the Highlands Highway in Papua New Guinea caused a rise in Fatal and Severe accidents from 3 in 7 years to 10 in 4 years. Conclusions

a) There have been very few studies indeed into the effects of geometric design upon accident rates in Developing Countries and it has taken a major effort in the US and Europe over some 30 years before regression equations of any substance have begun to emerge. This study has again shown the difficulties of such research, both in obtaining sufficient accident and survey data of adequate quality, and in analysing the data and revealing the sometimes complex nature of the relationships which may exist.

Notwithstanding these difficulties, some new and potentially valuable insights were gained in the study. The Papua New Guinea Highlands Highway data set proved to be the most fruitful, but this may have been due to the possibly rare circumstance that 13 years of accident data were available of relatively good quality, for a road that had had few changes over that period.

b) A digital video technique was adopted in an attempt to optimise the data collection phase of the research programme. It must be considered to have been only a partial success. Its use in the research plan depended upon the availability of engineering drawings for the highways. Whilst as-built drawings were obtained for the two highway sections studied in Nepal, the team was disappointed to only obtain them for 10km of road India and 10km in Zimbabwe.

This suggests that a combination of both PNG and video techniques is required in future studies. c) For accident investigators, the accurate location of accidents is essential in understanding the causes of the accidents. However, analysing the data suggested that the accuracy of police in locating accidents was in practice probably 1-2 km or more. It is recommended that in future analyses, values of geometric and other parameters assessed from the video should be estimated over 500m or 1km sections rather than the 100m adopted here.

d) For the Highlands Highway study, accidents were located to within a road section, which varied in length from 1 km to 20km, with an average value of 5 kms. With survey readings normally taken every 1 km, there were therefore 5 survey points per highway section over which parameters were averaged or the worst value found.

Out the outset of 5 the analysis, this limited resolution was considered to be a disadvantage, but the results suggest that in fact it was an adequate compromise given the volume of traffic (~1000 vpd), the number of accidents (5000 over 13 years) and the time taken to take measurements at each survey point (the 480 survey points took approximately 2 weeks to complete). e) It is recommended that appropriate speed measurement techniques, possibly using number plates, be developed and tested in any subsequent phase to this work.

f) Very good correspondence was found at two sites between predicted values of traffic volumes using the Wardrop and Charlesworth technique with video footage and actual traffic volume survey values.