We study a non-trivial extreme case of the orchard problem for 12 pseudolines and we provide a complete classification of pseudoline arrangements having 19 triple points and 9 double points. We have also classified those that can be realized with straight lines. They include new examples different from the known example of Böröczky. Since Melchior's inequality also holds for arrangements of pseudolines, we are able to deduce that some combinatorial point-line configurations cannot be realized using pseudolines. In particular, this gives a negative answer to one of Grünbaum's problems. We formulate some open problems which involve our new examples of line arrangements.Keywords line arrangements, pseudoline arrangements, orchard problem, Sylvester's problem1 The Sylvester-Gallai problemWe begin with a few definitions. A line arrangement in the real projective plane P 2 R is a finite set of lines in P 2 R . A pseudoline in P 2 R is a simple closed curve such that its removal does not cut P 2 R in two connected components. A pseudoline arrangement is a set of pseudolines in P 2 R such that every pair of pseudolines has precisely one point in common where the two curves intersect each other. A first book about pseudoline arrangements was written by Grünbaum [13]. It has turned out later that pseudoline arrangements are isomorphic to reorientation classes of oriented matroids in rank 3. This implies a close connection of our investigation to the theory of oriented matroids. The interested reader can find more about this relation in [2] and [3]. For a given line arrangement, or for a given pseudoline arrangement, we count the number of points that are incident with precisely k lines or k pseudolines, respectively, with k 2, and we denote this number of the arrangement by t k . We call a point incident with precisely r lines or with precisely r pseudolines an r-point. We use also the notion double point for r = 2, triple point for r = 3, and quadruple point for r = 4. We speak of an essential line (pseudoline) arrangement when all lines (pseudolines) do not intersect at one point. Our article can also be seen in the spirit of Grünbaum's book about point-line configurations, see [14]. The reader will get some benefit for understanding our paper when she/he has a look at this book. For a point-line configuration we have not only a set of lines but also a set of points together with an incidence relation between the set of points and the set of lines. It is clear that the lines can be replaced with pseudolines and we arrive at a point-pseudoline arrangement. Moreover, when we forget about any underlying geometric set of points, lines, or pseudolines, we arrive at an abstract point-line configuration. Point-line configurations with n lines and n points in which (•) the lines are incident with precisely k points and (••) the points are incident with precisely k lines have been