“The first results will be appearing shortly,” an announcer told the 40-odd journalists in the official election press-centre in Ceyhun Atuf Kansu Caddesi, Ankara.

“The first results will be appearing shortly,” an announcer told the 40-odd journalists in the official election press-centre in Ceyhun Atuf Kansu Caddesi, Ankara.

It was 5:45 pm. Polling had finished 45 minutes ago. Our laptops were open, and we were drinking tea and watching the big screen for the results from eastern Turkey _ where polling stations close before the rest of the country. Fifteen minutes passed, half an hour …and still the screen was unchanged. That was strange. I had witnessed seven elections and referendums in Turkey. Invariably we had the first results by 6 pm and the screen was changing rapidly by 6:15 pm.

By 6:30 pm the dearth of results had become a talking point. Something was going on. Then a message flashed on to the screen: The first results would appear at 6:45 pm.

It was the first of many surprises that night and, as with the others, it was troubling one. If one delays releasing results, one does it for a reason.

The campaign had generated excitement, particularly the punchy speeches of newcomer Muharrem Ince who drew crowds the like of which Turkey had not seen for many years … 2 million in Izmir, 1 million in Ankara (where Erdogan had drawn only 15,000), and at least 1.5 million in Istanbul.

“There was a huge surge in efficacy. People finally felt that their vote was going to matter this time,” Yasemin Acar, assistant professor of political psychology at Ozyegin University in Istanbul, was to tell me later.

I encountered the same confidence while interviewing voters on Sunday morning. “This time I believe we will have change,” my translator, a long-time opposition supporter, told me as we drove away from one polling station.

It was not to be. President Recep Tayyip Erdogan received 52.6 percent _ a figure higher than he had ever received, and he was certainly not more popular than he had ever been in his 15 years in power. Ince got 30.6 percent.

The word on opposition lips was “fraud”. Cumhuriyet reported that during the day police arrested three men in Sanliurfa province with four sacks of already-stamped ballots. A video circulated on social media that showed an electoral official stamping one ballot after another for Erdogan and his AKP.

But there are limits to such cheating when the electorate is as big as 59 million people.

Political scientists believe that Erdogan’s core vote _ those who support him whatever he does _ is about 3 5 percent of the electorate. On Sunday he got 17.6 percent more than that.

“Fraud could effect a difference of three to four percent, but a difference of 17.6 percent is impossible,” said Dogu Ergil, retired professor of politics from Ankara University.

The next day Ince made the same point. “Did they steal votes? Yes, they did. But did they steal 10 million votes? No,” he said in his concession speech. “There is no big gap between the Supreme Election Board data and the data that we collected ourselves.”

Many people were sceptical. Rumours circulated that as the results came in, Erdogan had called Ince to tell him not to push his success because there would be violence in the streets. Even the general of Turkey’s First Army had called Ince, the rumours said.

The next day Ince said he got no call from Erdogan or anybody else. If he had received such a message, he would have rebuffed it with contempt, he told the press.

I approached an old friend who is an expert on electoral fraud. Burcu Akcaru used to be one of the directors of Ankara Oylari, a civil society group allied with Oy ve Otesi which monitored the 2015 elections and helped to ensure they were clean.

Akcaru is now an executive of Iyi Party in charge of election security. IP had strong reason to suspect Sunday’s results because its score, and that of its leader Meral Aksener, were way below what had been forecast.

There were irregularities, Akcaru said, but these were “limited to certain regions. There was no evidence of big-scale fraud to affect the results”.

“The (Iyi Party’s) result-sheets coming from the polling stations are mostly matching with the results announced by the Supreme Election Board,” she said.

Akcaru pointed out that people find the election results hard to accept because they are so far from what polling companies had predicted and from what party workers had encountered in their door-to-door canvassing.

The director of Metropoll polling company, the only one to predict the AKP would lose its majority in June 2015, wrote something prescient in his June survey. “The final word,” Ozer Sencar said, “is in the hands of voters who either declined to express their opinion to us or who did not tell the truth when they did.”

Fear hung over the electorate like a dark cloud. The first person I approached outside a polling station on Sunday morning was a good-looking young man in a smart polo shirt in Aydinlik, Ankara. He declined to answer any question, saying “this is not a country of freedom, unfortunately”.

One of Sunday’s results did chime with what had been predicted: the AKP got 42.6 percent of the parliamentary vote.

The Turkish economist Ali Akarca has built an econometric model at the University of Illinois in Chicago which he uses to predict election results. The model is based on the results of 33 previous elections. He updates it with data such as per capital GDP, inflation rate, etc.

Akarca’s model predicted the AKP would get 42.8 percent. Metropoll forecast it would get 42.1 percent.

The main issue in the election was the economy. When Metropoll asked people “what do you think is Turkey’s most important problem?” nearly 54 percent said the economy and unemployment. After all, inflation was 12.5 percent, unemployment was over 10 percent, and the lira had fallen 25 percent against the dollar in the past six months.

This was fertile ground for the opposition. In his speeches, Ince repeatedly highlighted the government’s economic failings. Analysts thought the AKP’s vote could fall as low as the 40.9 it got in June 2015 _ when the poor economy was the main issue and AKP lost its majority.

On Sunday, the AKP again lost its majority _ but only by five seats. But what was baffling, if not bizarre, was that Erdogan increased his vote. The person who is most associated with the economy had profited from its turning sour.

“The negative developments in the economy,” Ozer Sencar told me in May, “have not yet affected people to the extent that it would have an impact on the elections.”

Professor Ergil gave another reason. Erdogan had convinced the voters that “it is not the Turkish economy that is in crisis, that it is the sinister foreign powers who want to put Turkey into a bad position.”

I had encountered this at polling stations. AKP voter Murat Conker, a tyre serviceman, said the bad economic data were due to “manipulation by external powers”.

My last task on election night was to give a live stand-up on the pavement of Eskisehir Yolu for France24 TV. We chose the position so the viewers would see, behind the reporter, cars of AKP supporters streaming along the highway, waving flags and singing Erdogan songs.

As my French-speaking colleague was speaking to the camera, a few AKP supporters came up and held flags of Turkey and the Ottoman Empire behind her, blocking out the background. Gently I tried to shoo them away.

“We are Ottomans,” one flag-holder said to me proudly. Fortunately he was not aggressive, but his fantasizing made me think of the basic unfairness of the election.

The government refused a head-to-head debate among the presidential candidates, it refused to grant equal time on state TV _ essential elements of elections in South Africa (where I come from) and other countries _ because the cut-and-thrust of such competition would have destroyed the fantasies of foreign powers manipulating the economy, their being no Kurdish problem, the Turkish Republic’s being just a pause in the march of Ottoman history, and so on.