Principal Bonginkosi Gumede is not keen to admit it, but he is all out of ideas. Too many girls in his rural KwaZulu-Natal school do not complete their education – and after eight years of trying various interventions, he now expects that to be the case this year, and the next, and probably the year after that.
"We have lots of girls who fall pregnant here," says Gumede. "Some drop out for six months, then they come back, but most do not come back. They become mothers, they stay at home, they never learn more."
In 2012, Gumede's school had 814 pupils. Seventy of them were pregnant during the year. Of those, 26 were under the age of consent, 16 years.
This school is not unique, nor is it an exceptional case, an annual survey of schools the department of basic education conducted in March, but has not yet released, suggests.
However, a Mail & Guardian analysis of a leaked version of the survey's raw data shows there are schools with higher numbers of pregnancies than Gumede's, larger ratios of pregnant pupils, more underage girls who fall pregnant and even younger children (many under 13) who fall pregnant.
Like several of those schools, Gumede blames a range of factors: social pressure on young girls to find a man; weak health and social services; and religious communities that do not exert any pressure of their own.
The department's annual survey, essentially a census of all schools in the country, is usually released as a well-digested report years after the data is gathered: the most recent officially available version dates from 2010, and was released in May 2012.
The M&G's analysis of the raw data on tens of thousands of schools shows enormous disparities between the health and wellbeing of school children in different provinces.
Startlingly, the data show that, in 2012, pupils at schools in the Free State were three times more likely to die than their compatriots in the Western Cape, and also – perhaps not coincidentally – 12 times as likely to fall pregnant.
Comparing causes of death across provinces as ratios of the number of pupils in each region shows that pupils in the Free State are far more likely to die of illness or violence than children in the rest of the country.
The sparsely populated Northern Cape, on the other hand, emerges as the province where schoolchildren are most prone to committing suicide. Limpopo, which reported the lowest rate of death due to accidents, violence and suicide, was nonetheless edged out as the safest province overall by the Western Cape, thanks to its remarkably low number of deaths due to illness.
According to the data, a total of 29 996 pupils are known to have been pregnant during the course of 2012. But the reality is likely to be several times that number thanks to under-reporting from school principals and because many older girls drop out of school without disclosing the fact that they had fallen pregnant.
A full third of those school pregnancies were recorded in the Eastern Cape, where 514 out of every 100 000 pupils were pregnant during 2012, according to their schools. In the Free State, with its smaller school population, that rate was 534.
By contrast, only 87 out of every 100 000 pupils in Limpopo were reported to be pregnant during the year, and in the Western Cape the number stood at just 42.
Based on their grade level, which is a rough measure of age, about 25% of the girls who were reported as pregnant – nearly 7 700 – were younger than the legal age of consent.
Data from individual schools also shows that about 37% of all pupils in the Eastern Cape receive social grants, as do about a third of all pupils in the Free State, the Northern Cape, Limpopo, Mpumalanga and the North West.
In the country's richest province, Gauteng, 13.3% of pupils received social grants. But in KwaZulu-Natal, the province with the largest population of schoolgoers, only 6.6% of pupils were recorded as receiving such grants.
The department of basic education has not released a report on the findings of the 2011 survey yet, making year-on-year comparisons impossible. Also not yet available is a final tally of how many schools submitted data, which would give some indication of the likely accuracy of the survey. However, numbers for both pregnancies and deaths are sharply lower than those released for 2010.
"We live in a society where a majority of young children are growing up without proper access to nutrition and healthcare," said Doron Isaacs, deputy general secretary of rights organisation Equal Education.
"It's also a society where there's a lot of crime [mostly affecting townships and other poor communities], and the state of police and the criminal justice system are not helping," he said.
"[Many] schools are abandoned after 2pm or 3pm. There's no school sport or any other extramural activities. Joining a gang becomes a sport for some young people."
Human rights lawyer Faranaaz Veriava said: "Current school pregnancy guidelines are ambiguous and have not been effective."
In response to questions, the department of basic education said that as a matter of principle, it "refuses to legitimise leaked documents with a comment – especially incorrect documents that are deliberately planted in the public space".
"However, we strongly refute any suggestion that, we have not released the information for public consumption. The data collated is used almost on a daily basis and for planning purposes as these issues are mainly societal matters that reflect on the broader society rather than the education sector alone." – Additional reporting by Bongani Nkosi