Dear experts,

I have a question concerning the handling of over- and underflow when using a *RooDataHist*. My ROOT version is 6.18, but I also tested the behaviour with 6.20.

I am importing a TH2D histogram via *RooDataHist* and plot the projection of the *RooDataHist* on the x axis.

It seems to me like this projection does not account for over- and underlow bins that are present in the initial TH2D.

To illustrate this behaviour, I can alternatively first create a projection of the TH2D on the x axis via the method *Projectionx()*, giving me a TH1D which fully accounts for the over- and underflow. I then import this TH1D via *RooDataHist* and compare it to the first projection.

Please find a simplified test example below:

```
TH2D *h2 = new TH2D("h2", "", 2, 0, 2, 2, 0, 2);
h2->Fill(0.5, 0.5);
h2->Fill(1.5, 1.5);
h2->Fill(0.5, -1);
h2->Fill(0.5, -1);
RooRealVar x("x", "x", -10, 10);
RooRealVar y("y", "y", -10, 10);
RooDataHist *dataHist = new RooDataHist("dataHist", "dataHist", RooArgList(x, y), h2);
TH1D *h1_x = h2->ProjectionX("projX");
RooDataHist *dataHist_x = new RooDataHist("dataHist_x", "dataHist_x", x, h1_x);
RooPlot* frame = x.frame();
dataHist->plotOn(frame);
dataHist_x->plotOn(frame, LineColor(kRed+2));
frame->Draw();
```

The 2 points in the x-y space at (0.5, -1) are not visible in the projection of the 2D *RooDataHist*.

Do I need to specify an additional option when calling *plotOn()* to account for the over- and underflow? Can I assume that the over- and underflow is in principle correctly accounted for when importing a 2D histogram through *RooDataHist*?

Your help is highly appreciated.

Best regards,

Philipp