On 6^th September 1986, Doctor Who returned to BBC1 in a blaze of publicity.  There were trailers, newspaper articles, an appearance on Wogan, Colin Baker opening shops around the country, and even a tie-in promotion with Golden Wonder crisps.

Doctor Who was at the heart of BBC1’s new Saturday night line-up between Roland Rat – poached from ITV at great expense – and Noel Edmonds. The first ever episode of Casualty was shown on the same evening. After the infamous 18-month gap, our favourite Time Lord was back. Actually, the gap was 17 months and one week to be precise. This article is about numbers and accuracy. You may want to hide behind the sofa now.

As we all know, the big launch didn’t end well. By the fourth week, only 3.7 million viewers tuned in to see the conclusion of the first segment of The Trial of a Time Lord. Colin Baker was out of a job shortly afterwards and Doctor Who didn’t return to Saturday schedules for nearly 20 years. 

One thing has bugged me about the story of the audience collapse of Season 23, and indeed the problems the series suffered until it was cancelled in 1989; the poll positions didn’t make sense. 

You may not know what a poll position is, so let me explain. It’s basically like the music charts but for TV shows. The most watched TV show of the week is in the number one spot, the second in number two, and so on.

What confused me was the poll positions for Season 23 and then 24 through to 26. On 6^th September 1986, according to every source I have seen, The Trial of a Time Lord Part One had an audience of 4.9 million and a poll position of 69. But Timelash Part Two had an audience of 7.4 million and only hit 79.  Mark of the Rani Part One got 6.3 million and was 111^th most watched show for the week. You can see the numbers yourself. It didn’t make sense.

A number of explanations have been offered for this statistical anomaly. TV audiences had declined between 1985 and 1986 due to rising competition from new-fangled video recorders, some said. Well, a bit, but not that much. TV audiences are lower in September than they are in winter. Also true, but not by enough and the same pattern shows when The Trial of a Time Lord Part Fourteen gets 5.5 million and 80^th place in December 1986.

So, I took myself to the BFI Reuben Library in London. In their comfortable surroundings, anyone can order up the British Audience Research Bureau reports for the years 1986 – 1989. I looked at the very weighty volume including week commencing 6^th September 1986. I took my pencil and ran it along the channels for that week (remember there were only four in 1986). I discovered that The Trial of a Time Lord Part One was not the 69^th most watched programme that week; it was 112^th.

Being the thorough person I am, I worked back. I did the same thing for Revelation of the Daleks Part Two. All the sources showed it had a poll position of 58 and my counting gave me exactly the same number. I tried some more episodes from 1983 through to 1985 and I found no discrepancies in the accepted numbers before 1986. I then tried Battlefield Part One, the lowest-rated episode ever with 3.1 million viewers. All the sources show a poll position of 102. My count showed it in 183^rd place for the week.  Much lower than thought. I tried a few more episodes, Seasons 23 through 26, and found the same thing, so I calculated them all…

Episode	Poll Position (believed)	Poll Position (actual)
The Trial of a Time Lord Part One	69	112
The Trial of a Time Lord Part Two	75	121
The Trial of a Time Lord Part Three	98	149
The Trial of a Time Lord Part Four	97	147
The Trial of a Time Lord Part Five	76	122
The Trial of a Time Lord Part Six	87	132
The Trial of a Time Lord Part Seven	87	137
The Trial of a Time Lord Part Eight	84	133
The Trial of a Time Lord Part Nine	85	136
The Trial of a Time Lord Part Ten	93	146
The Trial of a Time Lord Part Eleven	86	139
The Trial of a Time Lord Part Twelve	89	148
The Trial of a Time Lord Part Thirteen	98	158
The Trial of a Time Lord Part Fourteen	80	129
Time and the Rani Part One	71	105
Time and the Rani Part Two	85	130
Time and the Rani Part Three	81	126
Time and the Rani Part Four	86	121
Paradise Towers Part One	88	132
Paradise Towers Part Two	84	127
Paradise Towers Part Three	79	125
Paradise Towers Part Four	93	148
Delta and the Bannermen Part One	90	132
Delta and the Bannermen Part Two	93	145
Delta and the Bannermen Part Three	87	138
Dragonfire Part One	80	135
Dragonfire Part Two	96	145
Dragonfire Part Three	94	155
Remembrance of the Daleks Part One	78	128
Remembrance of the Daleks Part Two	78	120
Remembrance of the Daleks Part Three	91	144
Remembrance of the Daleks Part Four	96	147
The Happiness Patrol Part One	96	146
The Happiness Patrol Part Two	104	158
The Happiness Patrol Part Three	88	134
Silver Nemesis Part One	76	125
Silver Nemesis Part Two	94	147
Silver Nemesis Part Three	98	146
The Greatest Show in the Galaxy Part One	86	139
The Greatest Show in the Galaxy Part Two	99	154
The Greatest Show in the Galaxy Part Three	108	151
The Greatest Show in the Galaxy Part Four	79	124
Battlefield Part One	102	183
Battlefield Part Two	91	153
Battlefield Part Three	95	158
Battlefield Part Four	89	148
Ghost Light Part One	94	144
Ghost Light Part Two	93	153
Ghost Light Part Three	104	165
The Curse of Fenric Part One	104	166
The Curse of Fenric Part Two	104	169
The Curse of Fenric Part Three	104	174
The Curse of Fenric Part Four	104	173
Survival Part One	89	138
Survival Part Two	96	150
Survival Part Three	91	146

So why the discrepancy between poll positions for episodes before 1986 and those that came later?

Apparently, the source for the ratings information we use today was the Doctor Who production office in the 1980s. The reason for the difference is, someone has used a method of counting that significantly favours the later group of episodes. 

Up to Revelation of the Daleks, the poll position is calculated by counting every programme shown in a given week. So, for example, if News at Ten is shown five times in a week, then it occupies five places. If Crossroads is shown three times, then it occupies three places, and so on. From The Trial of a Time Lord, only the show is counted, so News at Ten occupies one place and Crossroads occupies one place. As a result, the poll position for any programme with a rating lower than News at Ten or Crossroads improves.

This matters increasingly the lower your audience is. For Revelation of the Daleks Part Two, with an audience of 7.7 million, the poll position rises from the generally acknowledged 58 to 39 if you count it the second way. For Battlefield Part One, the difference is a whopping 81 places. 

The reason it matters more for Battlefield Part One is that, when your programme only gets an audience of 3.1 million, you come in behind pretty much everything — One O’Clock News, early afternoon kids’ shows, late afternoon games: you name it. 

So why did this change happen?  It’s tempting to picture John Nathan-Turner, perhaps with a small stiffener in hand, faced with the failure of Season 23, coming up with a creative way of casting the figures in a better light. Who can say?