The Times May 2013 A journal of transport timetable history and analysis

Measuring the world’s worst timetable performance

RRP $4.95 Inside: “Central” versus “” Incl. GST NZR conditional trains in its PTT What’s in a name-4? How bad can a timetable get? The Times Journal of the Australian Association of Time Table Collectors Inc. (A0043673H) Print Publication No: 349069/00070, ISSN 0813-6327 May 2013 Issue No. 351 Vol 30 No. 05

—Contents— LETTER– NSW LINES SERVED BY PASSENGER TRAINS 3 CONDITIONAL TRAINS IN A PUBLIC TIMETABLE 4 WHAT’S IN A NAME (4)? 8 HORSE KICKS AND LATE TRAINS 13

On the front cover

Railways and railway timetables have always been fair game for humorists and cartoonist. The cartoons of Roland Emmet (mostly affectionate cartoons of the Far Tottering and Oyster Creek Railway) were best sellers in book form and are true collec- tors’ items. The Southwold Railway became the butt of some rather less affectionate cartoons .by Reginald Carter. Carter made a name for himself as a comic picture postcard artist, his first commission being at the age of 17. He had a studio built at the bottom of his garden and started publishing his best-known postcard series: The Southwold Railway Cartoons. During the First War, he produced many more cards gently lampooning army life. A selection of his postcards on the Southwold can be found at http://transportsofdelight.smugmug.com/RAILWAYS/BRITISH-STEAM-LOCOMOTIVES/NARROW-GAUGE/17391786_3vvzc9/1546653359_ZRDhcd6#! i=1546653359&k=ZRDhcd6 A Southwold Railway Timetable from the July 1922 Bradshaw shows a service of 6 slow trains per day over this little line. Was the service as bad as that currently provided on V/Lines Albury line today? This question is examined within the pages of this issue of The Times.

Contributors Robert Henderson (2), Victor Isaacs, Geoff Lambert, Reg Carter The Times welcomes articles and letters. Send paper manuscripts or word-processor files on disk or via e-mail to the editor at the address below. Illustrations should be submitted as clean sharp photocopies on white paper or scanned GIF or TIF format images with at least 300 dpi resolution on disk or via e-mail. Reproduction Material appearing in The Times or Table Talk may be reproduced in other publications, if acknowledgment is made. Disclaimer Opinions expressed in The Times are not necessarily those of the Association or its members. We welcome a broad range of views on timetabling matters. The Times on-line AATTC's home page: http://www.aattc.org.au has colour PDF versions of The Times

President Victor Isaacs 43 Lowanna Street BRADDON ACT 2612 [email protected] Secretary Michael Smith [email protected] Editor, The Times Geoff Lambert 179 Sydney Rd FAIRLIGHT 2094 NSW [email protected] Editors, Table Talk Craig Halsall 27 Edithvale Rd EDITHVALE VIC 3196 [email protected] Victor Isaacs 43 Lowanna Street BRADDON ACT 2612 [email protected] Distribution Officer Len Regan PO Box 21 YEA VIC 3717 0409 209114 [email protected] Membership Officer Dennis McLean P.O. Box 1253 NORTH LAKES 4503 (07) 3491 3734 Webmaster Lourie Smit [email protected] (02) 9527 6636 Adelaide Convenor Roger Wheaton 2C Bakewell Street, TUSMORE SA 5065 (08) 8331 9043 Canberra Convenor David Cranney Brisbane Convenor Brian Webber 8 Coachwood St KEPERA Qld 4054 (07) 3354 2140 Melbourne Convenor David Hennell (03) 983 01802 Sydney Convenor Geoffrey Clifton PO Box 5062 OLD NSW 2146 0405 387478 2 The Times May 2013 Le t t e r – NSW lines served by passenger trains from ROBERT HENDERSON

am writing to comment on the list of 1960s, but one gets the impression that the Radiating from this point (or these passenger train lines by the Master of diagrams may have been drawn many points?) in the list are “Central-St James” I the Lists, Ross Willson, in the April years prior to that. So hopefully it can be and “Central-North Sydney-Hornsby”, issue of The Times. taken as a reference document as at the but “Sydney-Bourke”. So the question The year 1943 is significant for both the date of our list. Looking at my well pre- arises in my mind as to how, in terms of Master and this correspondent, so it was served 1960s-bought masterpiece, I could the list, a suburban electric train arriving very timely that such a list be published. not help but note these specific junction at Central from St James continued its However, I wonder if time is catching up names, not recorded in the list: journey to (say) Cronulla? Did it follow the “Sydney-Bourke” line for the short with some of us. So firstly I point out that, Clyde Junction (as junction of the Car- distance to Illawarra Junction? Page 3 of while the title of the list clearly states lingford line) “Lines … within New South Wales”, men- our prime reference document shows that tion is made of lines to South Brisbane, Goulburn North Junction (Crookwell) Sydney and Central are separate stations (with a height differential of four feet!), Canberra and Wodonga! Wallerawang West Junction (Gwabegar) and that different curves and gradients Secondly, for some lines, very specific Nyngan Junction (Cobar) apply to the subject section of line. [see junction names are recorded in the list. My below, Editor] perception of the nomenclature used in the East Greta Junction (Cessnock) And, finally, a question seeking informa- list is that reference was possibly made to South Junction (but presumably Narrabri tion. As 1943 was at about the height of the marvellous New South Wales Rail- South Junction in full) (Walgett). ways’ Curve and Gradient Diagrams. My World War II, did troop trains (which copy was purchased sometime in the early Thirdly, a matter for debate is whether surely carried passengers) travel via the Sydney is the same station as Central. North Strathfield Loop?

The two Sydneys. A long time ago, I realised that even taxi drivers knew the difference between “Central” and “Sydney”- more correctly “Sydney Terminal”. This is unusual in this day and age as most taxi drivers would not know their knee from their elbow. These names persist even in current maps and track diagrams, as shown by the extracts from the current Drivers’ Route Knowledge maps. The 0 km post is shown for Central, but not for Sydney Terminal, although I believe it is meant to be at the Platform 1 buffer stops.– Geoff Lambert, Editor.

The Times May 2013 3 Conditional trains in a Public timetable: New Zealand Railways 1951 VICTOR ISAACS

ew Zealand Railways issued their significant impact on society. NZR was thrice weekly “during summer holiday booklet of “Time-Tables North still recovering from post-war coal and season only” but the dates are not given. Island on and after June 3, 1951”, materials shortages. Thirdly – and most N What is also interesting is that provision is new series, no. 2. This was the period unusually – is the setting out in detail in a when separate Public timetable booklets Public timetable, the timetable of a condi- also made in this Public timetable for con- were issued for the North and South Is- tional train. This is designated “Relief sequential alterations to other services. lands. The Public timetables not only in- Express”. The footnote says this “Runs Some of the very local trains have side cluded trains, but also – at the back - NZR only when required”. Presumably this notes indicating they have different times bus services, of which, in this era, there would be at holiday times, but this is not in the event that one of the conditional were many, both rural and suburban. stated. extras runs. For example, a note states that no. 592, the 4.15 pm from Taumaranui to Our first illustration is of the line to North- The next two illustrations are of the most Frankton Junction “Leaves Taumaranui at land. I think that there are a few interesting important railway line in New Zealand – 5.0 pm when J-10 special “Daylight Lim- aspects of this. First, is the fairly lavish the North Island Main Trunk (as New Zea- ited” express runs”. We can see three such provision of local trains. These are all landers call mainlines) from Auckland to consequential alterations in this double marked with an “S” symbol standing for Wellington. In these timetables we see that page spread. Second class only. In practice, these were there are the schedules for no less than very slow Mixed trains. Secondly, there is three conditional extra passenger trains. All of these consequential alterations were the meagre provision of long distance One was a relief to the overnight Express to local trains, of which in this period, trains. There was a reasonable train from and one a relief to the overnight Limited there were many on the Main Trunk. What Auckland to Whangerei, the principal town Express. If both of these ran, there would seems especially unusual to us today is the of the region, just thrice a week! This is the be four overnight trains on the length of provision of extra Mixed trains from some train designated “Express” and running the Main Trunk. The third possible addi- of the towns very late on Saturday nights. only on Mondays, Wednesdays and Fri- tional train could perhaps be called an These would have been to take the picture days on both directions. Unfortunately, this occasional, rather than a conditional train. goers home. Perhaps, however, there were is all too typical of NZR in this period – This was the Daylight Limited. This ran only a limited number of drunks, as 6 the time when motor cars were making a from Wellington to Auckland and v.v. o’clock closing was definitely the rule in NZ then.

4 The Times May 2013 The Times May 2013 5 6 The Times May 2013 The Times May 2013 7 What’s in a name-4? Named Sydney private bus routes– City Bus Direct ROBERT HENDERSON

ost bus routes are known solely in the north-western suburbs Routes 621 and 622 turned out to be rela- by their number and end destina- of Sydney. tively short-lived, as they both ceased on 8 tions. However, a small number October 2004, with the company saying M Curiously Route of Sydney private bus routes have also had that there had been insufficient patronage 620’s Dural terminus was almost directly names attached to them, even if only for to make them commercially viable in the outside the New Line Road depot of its short periods of time. This is the fourth in days when private bus operators relied rival in the area, Glenorie Bus Company. a series, briefly surveying those that spring entirely on fare revenue to maintain ser- City Bus Direct Route 620 and to mind. vices. Route 642 ran in competition with each CITY BUS DIRECT other along New Line Road and Pennant After 621 and 622 ceased, 620 continued Hills Road, through Dural, Cherrybrook to run. But within weeks the whole Harris The name “City Bus Direct” was a clever and West Pennant Hills. When this degree Park Transport network of bus services, use of the initials “CBD” for new bus of competition came to light, some influ- including Route 620, was transferred to routes running from a number of suburban ence was brought to bear to get Glenorie Group’s Hillsbuscom- areas into the centre of Sydney. The three Bus Company to alter its route so that, pany on 22 December 2004, as Harris Park City Bus Direct routes were another ven- from 8 July 2002, its buses ran along Transport management elected not to enter ture of Harris Park Transport. They bore streets other than New Line Road for a into a contract with the Government after route numbers: considerable distance. amendments to the Passenger Transport Act. Their departure from running route 620, from the corner of Old Northern and Harris Park Transport was keen to present service buses marked the end of the “City New Line Roads (at the McDonalds store), its new routes as being operated by the Bus Direct” name. Dural to the City via Cherrybrook, which separate entity, City Bus Direct, as it set up started on 4 March 2002; a separate website in that name, complete Following Hillsbus’s acquisition of the ex- 621, from North to the City via with timetable and fare information. Both Harris Park routes, some drama occurred. North Rocks, Oakes and Aiken Roads, the website and paper timetables pro- While Hillsbus was able to run school which started on 11 June 2002; claimed that the service was being run with holiday services on its new routes during brand new, air conditioned Mercedes Benz January 2005, when the 2005 school year 622, from Oatlands to the City via Pennant buses. They erected J-poles at bus stops approached, Hillsbus apparently found Hills and Jenkins Rods, Dunrossil Street with the City Bus Direct logo, depicting a itself with insufficient resources to provide and Ray Road, which started on 22 April likeness of the Sydney Harbour Bridge. all school day services. So, after only five 2002. They also issued electronic credit-card- weeks of operation by Hillsbus, the routes sized debit cards, known as “CBD easy Each of the routes had a catchment area in were further transferred to Sydney Buses, cards” to speed up boarding and so that the already existing contract area of the operating out of its Ryde depot. Sydney customers received discounts of 15% on company, and then ran express along the Buses continued to run all the ex-Harris the normal fare, as well as getting their and across the Har- Park routes for another eight months. Friday trip free of they had travelled regu- bour Bridge to the Queen Victoria Build- larly on Monday to Thursday. Marketing During those eight months, in August ing, near Sydney Town Hall. They were of the “City Bus Direct” image was en- 2005, the management of both Hillsbus the first and only routes which Harris Park hanced by the display of those words on and changed from National Ex- Transport ran into the City. Their remain- destination displays for City-bound trips. press Group to the consortium of Comfort ing routes were all entirely suburban. DelGroCabcharge (CDC), which is owned Routes such as those labelled City Bus Route 620 commenced with ten morning partly by Singapore Bus Service and partly Direct were made possible when the Pas- trips to the city, leaving Dural between by the Australian-owned Cabcharge, a senger Transport Act was passed back in 5.57 and 9.20 am, and ten afternoon return business associated with taxi operations. 1990. The Act enabled bus operators to trips, with the first leaving the city at The new management was able to negoti- take passengers from within their contract 1.45pm and the remainder between 4.15 ate with the Government to retransfer areas to points outside, but without con- and 7.10 pm. From 9 September 2002, the many of the ex-Harris Park Transport veying those passengers totally outside the number of trips increased to 20 inwards routes, including Route 620, back to Hills- contract area. throughout the day and 18 return. The bus as from 25 September 2005. timetable remained at about that level for The first of the three City Bus Direct the rest of the time that the service was run Both Hillsbus and Sydney Buses main- routes (620) started only three weeks after by Harris Park Transport and then up till tained exactly the same timetable for Route Glenorie Bus Company had commenced a 2009. 620 as Harris Park Transport’s City Bus similar service numbered 642, from Round Direct from December 2004 until the Min- Corner, Dural to the City, on 11 February The other two routes ran less frequently. istry of Transport’s review of Region 4 2002. At the time, Glenorie Bus Company Route 621 had four trips each way and took effect from 11 May 2009. In that was owned by the British National Express Route 622 eight each way. In both cases review Route 620 was reduced to a peak Group and they branded that part of their the trips on these routes ran only in the hour service, becoming supplementary to operation as “Hillsbus”. Later on – in peak direction. During Harris Park’s con- the more regular, seven-day-a-week Route December 2004 – the name Hillsbus was trol, all three City Bus Direct routes ran 621, which has as its suburban terminus extended to the larger group of routes run solely on weekdays. the major commercial centre of Castle Hill, by Westbus which lay generally within the rather than McDonalds at Dural.

8 The Times May 2013 The Times May 2013 9 10 The Times May 2013 The Times May 2013 11 What the Prussian Army horse can teach us about V/Line train timetables GEOFF LAMBERT

adislaus Josephovich Bortkiewicz # of Total # with (1868 –1931, near right) was an N years (%) N deaths L economist and statistician of Polish 0 109 54% 0 descent, who lived most of his professional 1 65 87% 65 life in Germany. In 1898, while he was 2 23 98% 46 working as a statistician for the Russian 3 3 100% 9 Railways, he published a book titled The 4 1 100% 4 Law of Small Numbers. In this book he Sums 201 124 first noted that events with low frequency in a large population follow a Poisson distribution even when the probabilities of the events varied. It was that book that made the Prussian horse-kick data famous. His data gave the number of soldiers killed by horse-kick each year in each of 14 cav- alry corps over a 20-year period. There were 201 “Corps years” in his database. He found that deaths by horse kick fol- lowed the pattern shown in the Table at upper right. A chart of these data appears below.

The total deaths from horse kicks were 124, and the average number of deaths per year per corps was thus 124/201 = 0.62. This is a rate of less than 1. In any given year, we expect to observe— well, not exactly 0.62 deaths in one corps (that is not possible because deaths occur in modules of 1)— but sometimes (in this case 54% of a discrete probability distribution that ex- the time) none, sometimes one, occasion- presses the probability of a given number ally two, perhaps once in a while three, and of events occurring in a fixed interval of (we might intuitively expect) very rarely time and/or space if these events occur any more. In this study there never was a with a known average rate and independ- year in which 5 or more deaths occurred. ently of the time since the last event. The The black line shows how the percentage Poisson distribution can also be used for of deaths climbs as one increases N, the the number of events in other specified number of deaths per year. Bortkiewicz intervals such as distance, area or volume. showed that those numbers follow a Pois- For instance, suppose someone typically son distribution. The Poisson Distribution gets 4 pieces of mail per day on average. is named for its discoverer (above, far There will be, however, a certain spread: right), who first applied it to the delibera- sometimes a little more, sometimes a little tions of juries; in that form it did not attract less, once in a while nothing at all. Given wide attention. More suggestive was Pois- only the average rate, for a certain period son's application to the science of artillery. of observation (pieces of mail per day, phone calls per hour, etc.), and assuming Wikipedia says: the Poisson distribution is that the process, or mix of processes, that produces the event flow is essentially ran- dom, the Poisson distribution specifies

# of "Corp-years" with N Deaths by horse kick how likely it is that the count will be 3, or 5, or 11, or any other number, during one 120 100% period of observation. That is, it predicts 90% 100 the degree of spread around a known aver- 80% # of corp-years age rate of occurrence. 70% 80 Cumulative # of deaths as %age 60% A discrete stochastic variable X is said to 60 50% have a Poisson distribution with parameter 40% 40 λ > 0, if for k = 0, 1, 2, ... the probability 30% mass function of X is given by: 20% 20 10% 0 0% 12345 Deaths per year (N) where e is the base of the natural logarithm 12 The Times May 2013 Number of trains Target transit performance 600 0 to 11 mins late 543 trains

500 11 to22 minutes late 33 trains 400

300 22 to 33 minutes late 1 train 200 lates to what happens out on the track. We 100 are particularly concerned here with this 0 statistic because V/Line constantly blames -11 to 0 0 to 11 11 to 22 22 to 33 33 to 44 44 to 55 55 to 66 66 to 77 77 to 88 88 to 99 99 to 110 110 to 121 121 to 132 events that happen in transit over ARTC’s Transit lateness range (minutes) track for most of its woes. Number of trains Actual Transit Performance 200 We should note before proceeding further 180 that 96% of timetabled trains are supposed 160 140 to actually run. On the NESG, V/Line 120 schedules 6 trains per day or about 183 per 100 80 month. Ninety-six percent of this is 175 so, 60 in theory, we would regard anything less 40 20 than this as “poor performance.” The real- 0 ity on the NESG is not just a poor perform- -11 to 0 0 to 11 11 to 22 22 to 33 33 to 44 44 to 55 55 to 66 66 to 77 77 to 88 88 to 99 99 to 110 110 to 121 121 to 132 ance, but a woeful one. In March 2013, for Transit lateness range (minutes) instance, only 23% of trains actually ran Percentage of trains this late Transit lateness lines against a target of 96%. While this low figure is terrible in itself, it also causes 120% trouble for the statistician who has many 100% fewer trains to analyse for lateness. There

80% is some hint that V/Line prefers to cancel

60% trains rather than reporting them as running Actual late. 40% 92% level 20% Target So, for V/Line, what does the Poisson

0% distribution predict for how the trains -15 -10 -5 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 ought to run? We ask the following ques- Minites late tions: “How many trains run no later than 1 (e = 2.71828...) and; k! is the factorial of k long-distance trains, its web-site says: lateness unit or 11 minutes?; how many and; λ=λT when the number of events run 11 to 22 minutes late?; how many run Our target is for at least 96% of sched- 22 to 33 minutes late? – and so on. We can occurring will be observed in the time uled services to run interval T=1. The positive real number λ is use Poisson’s equation (above) to answer equal to the expected value of X and also Our on time target is 92% this question and apply it to the number of to its variance. Clear? Good! On short distance, commuter services trains in my database. The database covers punctuality is measured on time to 5 the 8.5 months from August 2012 to mid- The horse-kick data, then, is the classic minutes 59 seconds. April 2013. There are 577 trains for which Poisson situation: a rare event, whose aver- I have records of running times. age rate is small, with observations made On long distance services it is measured over many small intervals of time. For a on time to 10 minutes 59 seconds. If V/Line were to meet its stated target, the Poisson distribution the variability (more statistics ought to pan out as shown in the V/Line has used a performance target of Table above. And the chart ought to look particularly the “variance”) is equal to the 92% rather than the 95% commonly used average- in this case 0.62. The square root like the chart at top left. The third chart is a in statistics, but that doesn’t matter mathe- line version of the same chart with the of the variance is the statistic known as the matically. For its “units of time”, V/Line standard deviation and in this case it is target performance shown by the dotted has chosen 11 minutes, rather than the line and the 92% level by the red horizon- equal to 0.78. From statistical theory, we “corps-years” we chose for the horse-kick can say that an observation that is more tal line. On the horizontal axis below deaths. Although this is a somewhat un- where the red and dotted lines intersect is than two standard deviations away from usual number to pick (it is even more pecu- the average will occur only 5% of the time. the time at which 92% of trains should fall. liar because it is 10 minutes and 59 sec- The average “expected” lateness (akin to Such an observation would be classed as onds); again this does not affect the mathe- unusual or, in the lingo, “statistically sig- the average number of horse-kick deaths matics. Let’s call it a “lateness unit” in the per year) is 5 minutes. nificant”. Ninety-five percent of the time same manner as we referred to a “Corps- the observation will be closer to the aver- year”, above. What is the reality? It is horrible. The sec- age than this and we dare not say they are ond chart shows how horrible. In V/Line’s anything out of the ordinary. For Prussia’s It is unclear what the V/Line jargon “on real world, the average (median) lateness is horses this 95% level is 0.61 + 2*0.78 time” really means. One assumes it means 40 minutes, against a target of 5 minutes. =2.19 deaths per year. Indeed we see in “on time arrival at the final destination”. We have to stretch out our lateness crite- this case that more than 2 deaths per year Of course, trains may be late departing rion to about 63 minutes before we can occurred only 2% of the time. from their origin. The arrival lateness is cover the transit times of 92% of those 577 thus the departure lateness added to the Trains, Planes and Automobiles are like trains. The chart at top right on page 14 time lost on the journey. It has been shown shows what percentage of trains ran “late” this. More particularly, train performance in the transport and operations research against a train timetable is like this. In - i.e. 11 minutes or more late—over the 9 literature that departure times are not Pois- months to April 2013 setting Key Performance Indicators, a rail- son-distributed, but that arrival times and way enterprise will set a target of a certain transit times are. Both departure lateness The most common (modal, or the peak of proportion of trains to arrive within a par- and time lost in transit are of interest to the Chart 2) delay is somewhere in the 22 to ticular time window. In the case of V/Line analyst- the latter more so because it re- 33 minutes time slot– say about 27 min-

The Times May 2013 13 Minutes late utes. This number appears to have been Percentage of V/Line trains which ran late, latched onto by V/Line. In December Average lateness (minutes) 2012, V/Line published an extraordinary 100% 60 “delay timetable” (pp 15-16). In this time- 50 table it cautioned customers that trains 80% would “possibly” be delayed by about 25 40 minutes over the course of the NESG jour- 60% ney. The publication of such a timetable is 30 surely a world first and a frank admission 40% of defeat. V/Line feels safe in doing so %age late (left axis) 20 though because it sheets home the blame to Minutes late (right axis) ARTC. V/Line distributed those 25 min- 20% 10 utes evenly over the journey, and equally for all trains. There is some evidence that 0% 0 delays accumulate at particular choke Aug-12 Sep-12 Oct-12 Nov-12 Dec-12 Jan-13 Feb-13 Mar-13 Apr-13 points like the Passing Lanes and on lengths of speed-restricted track. There is also evidence that some trains are better performers than others— often because they have planned standing time “delays” at Passing Lanes. This allows them to “crib” time and cancel out accumulated days when no crossing is needed. ARTC has a somewhat different set of criteria for its tracks in general and for the NESG in particular, which it regards as part of the Sydney-Melbourne Corridor. There are very few freight trains that are truly “Sydney-Melbourne only”, so ARTC’s statistics are concerned primarily with passenger trains- which is what we are also interested in. Indeed passenger trains still form half of all the traffic over the NESG even when all other freights are included. ARTC does not analyse Riv- erina-Melbourne Freight services. 1000 # of services (pass. and freight) per quarter, NESG ARTC has set 15 minutes as its tolerance 900 800 for “lateness”. Performance against timeta- 700 ble fell from 85% a few years ago to 35% 600 in the third quarter of 2012 (chart middle 500 400 right, NESG arrowed). This is almost en- 300 tirely due to the phased introduction of the 200 new V/Line NESG services, as can be seen 100 in the third (derived) chart. These perform- 0 1 4 4 3 4 3 Q -Q2 Q Q1 Q -Q1 Q Q Q Q4 09-Q3 0-Q2 1-Q2 2-Q1 ance figures are better than those for V/ 008-Q4 010-Q3 012-Q2 Line, because the well-performing Coun- 2 2009- 2009 20 2009- 2010- 201 2 2010- 2011 201 2011- 2011- 201 2 2012- 2012- trylink trains form 40% of the passenger the punctuality only from the long term services on the NESG. 160 years of Australian railway history, average of 4% to about 70%- this is no probably in world history. The 3rd quarter of 2012 shows an improve- good for V/Line because it would continue ment in performance. This is due mostly to to incur fines for late-running. To get to The Southwold Railway the high rate of cancellation of V/Line 92% punctuality, we need to slow the time- My paradigm for this critique of V/Line’s trains. There are simply fewer trains on the table by 63-11 = 52 minutes, so that 92% NESG service is the Southwold Railway. line and thus fewer to run late. This too can of trains would be no more than 11 min- The Southwold Railway was a narrow be seen in the third chart. The first quarter utes late. This course of action has been gauge railway line between Halesworth of 2013 will surely continue this trend. talked about in discussions between ARTC and Southwold in the English county of and V/Line, but it has never happened. As How could V/Line fix these problems? Suffolk. 8¾ miles (14.1 km) long, it was was pointed out in an earlier article in this 3 ft (914 mm) gauge. It opened in 1879 series, this would have slowed down the There are two ways of answering this th and closed in 1929. It became the subject question (in the interim before track and trains to 19 Century performance. of ridicule because of its unreliability, lack rolling-stock repairs are made)– slow the 2. Move the goal posts of punctuality and general unfriendliness. trains or move the goal-posts. It gained this notoriety mostly from a We could change the criterion for young commercial artist, Reg Carter, who 1. Slow the trains “lateness” to 63 minutes. This would be worked from a studio on the line. What If we were to ignore the Poisson distribu- ridiculous, so it won’t happen. would he have done with V/Line? An evocative video of the last train can be tion of the delay times, we would be Taking in to account both the reliability found at http://www.youtube.com/watch? tempted to slow the trains by the 25 min- and punctuality figures over the best part v=ygtjCLh4YBs. utes which V/Line has conjured up in its of a year, one can fairly say that these delay timetable. A little reflection, how- trains are the worst performers in nearly ever, would show that this would increase 14 The Times May 2013 The Times May 2013 15 16 The Times May 2013