At the end of December we heard of course the bad news of the Forensic Science Service that closes in 2012 due to losses. Since it is the biggest provider of England and Wales (60 percent) owned by the government it is a question what happens with the experience of the forensic scientists. Of course I also heard the bad news from many colleagues who work there, and since they were very active within the forensic communities, such as ENFSI, we will also see changes in these law enforcement forensic communities. Perhaps we learn from what happened there, however perhaps future will give them also new opportunities.
Also in the Netherlands it is investigated what happens if we are working in a market as a forensic service. This report is balanced, since it also provides insight in the issues that can arise in a market. An abstract in English is provided on the website, and describes three scenarios from the full government service to the free market solution for a crime lab with strict market control with a regulator.
An issue with forensic science in a crime lab is that there is not many jobs, and certainly not if budget cuts are in place. There are many students interested in forensic science as such, also due to the CSI effect, so we see a mismatch between the number of jobs and the available jobs. If we take England and Wales as and example, the Forensic Science Service has 1300 scientists and figures state that it is 60 percent of the market. The population of England and Wales is 53 million, so it means that aproximately 40 scientist per million of inhibitants. If we assume that forensic scientist stay on their job on avarage for 20 years, this means 2 vacancies per million inhibitants. Most European countries have less forensic scientists, so the number of vacancies for crime labs is limited, at least in crime labs.
Sunday, December 26, 2010
Friday, November 26, 2010
December 2010
From the last posting on possible errors in forensic science, I received several reactions for methods to reduce errors caused by humans by software. Of course that is a good approach, however one should include it in the whole system.
One issue we also see is the way the conclusions and the report are interpreted by the readers such as judges, juries and prosecutors. Also the combination of conclusions from different sources of evidence is important. For this a Bayesian approach can also be used, which is logically correct. An issue remains that the readers not always read the same, and interpretation by the readers of the report can also give confusion. For that reason, it is important to have good education on statistics, and write clear reports. And there we have the paradox that by writing more logically reports it is less intuitive for the reader, especially in detailed technical reports.
Currently also working on my website http://forensic.to/forensic.html , since sometimes information has to be checked and rechecked again. It seems many persons also use older versions, and ask for updates of these. This week I had to stay at home due to an injury at my knee, so I had time to read articles, reviewing them, and also making phone calls. Working at the computer was less ergonomic. Next week I hope to start working again.
One issue we also see is the way the conclusions and the report are interpreted by the readers such as judges, juries and prosecutors. Also the combination of conclusions from different sources of evidence is important. For this a Bayesian approach can also be used, which is logically correct. An issue remains that the readers not always read the same, and interpretation by the readers of the report can also give confusion. For that reason, it is important to have good education on statistics, and write clear reports. And there we have the paradox that by writing more logically reports it is less intuitive for the reader, especially in detailed technical reports.
Currently also working on my website http://forensic.to/forensic.html , since sometimes information has to be checked and rechecked again. It seems many persons also use older versions, and ask for updates of these. This week I had to stay at home due to an injury at my knee, so I had time to read articles, reviewing them, and also making phone calls. Working at the computer was less ergonomic. Next week I hope to start working again.
Thursday, November 11, 2010
November 2010
November started with the organization of the Themadagen Digital Evidence for the Dutch Law Enforcement, were we had 270 participants. Furthermore as always a somewhat busy time, however this month some vacation time (since normally I work during the summer holidays). So going south to Cyprus, with nice weather.
Also thinking on forensic science for law enforcement and commercial services and issues like certification. In the Netherlands they are working on a register for certified forensic scientist, which is in the law.
There are several opinions on this development. An opinion is that forensic science for crime cases is not good as a commercial service, since it might cause issues with bias to the client who pays for the services. Others think this can be prevented within the law system by making judges, prosecutors and juries if applicable aware of this issue, and that it will also be solved within a market, as in the UK, with a forensic regulator by using appropiate scientific quality standards.
An issue which we see in England and Wales, is that the police will do more forensic work themselves, so there is no market anymore. In Germany and other countries it appears to work well with private experts.
However, with any system all people should be aware of objective forensic services, and for this reason all research for objectivation of forensic evidence is impportant, were now much effort is spend by the forensic community, in funding of research by the National Institute of Justice and of the FP7 of the European Union.
Also thinking on forensic science for law enforcement and commercial services and issues like certification. In the Netherlands they are working on a register for certified forensic scientist, which is in the law.
There are several opinions on this development. An opinion is that forensic science for crime cases is not good as a commercial service, since it might cause issues with bias to the client who pays for the services. Others think this can be prevented within the law system by making judges, prosecutors and juries if applicable aware of this issue, and that it will also be solved within a market, as in the UK, with a forensic regulator by using appropiate scientific quality standards.
An issue which we see in England and Wales, is that the police will do more forensic work themselves, so there is no market anymore. In Germany and other countries it appears to work well with private experts.
However, with any system all people should be aware of objective forensic services, and for this reason all research for objectivation of forensic evidence is impportant, were now much effort is spend by the forensic community, in funding of research by the National Institute of Justice and of the FP7 of the European Union.
Saturday, October 23, 2010
October 2010 (2)
This week I went to Dalian, China to BIT's 1st forensic conference . It was a very nice venue and there were several tracks in parrellel. The opening ceremony was indeed impressing, also with Henry Lee, who told he was born in Taiwan, and also emphasised the importance of search strategies in forensic science.
Dr. Gerry LaPorte from the National Institute of Justice, emphasised also the culture within the forensic laboratories in the past that if you would make an error you could forget your carreer in forensic science. This is not good, since there is no room then to learn from past experiences, and improve the system. Also the implications of the NAS report will lead to more research to objectivation.
The conference had many good papers in objectivation, from speaker analysis to handwriting, cybercrime investigation to facial comparison and the limitations that exist there. It was good to see that there is more research on measuring the system and the error rate in forensic science, as well as having statistical analysis for improvement, also with the use of likelihood ratio calculations.
Dalian was also a very nice city to visit, near the waterfront, and with a fast developing center, with good transportation options, and excellent service. The confrence itself was also good for many new contacts within forensic science, and will also be organized next year.
Dr. Gerry LaPorte from the National Institute of Justice, emphasised also the culture within the forensic laboratories in the past that if you would make an error you could forget your carreer in forensic science. This is not good, since there is no room then to learn from past experiences, and improve the system. Also the implications of the NAS report will lead to more research to objectivation.
The conference had many good papers in objectivation, from speaker analysis to handwriting, cybercrime investigation to facial comparison and the limitations that exist there. It was good to see that there is more research on measuring the system and the error rate in forensic science, as well as having statistical analysis for improvement, also with the use of likelihood ratio calculations.
Dalian was also a very nice city to visit, near the waterfront, and with a fast developing center, with good transportation options, and excellent service. The confrence itself was also good for many new contacts within forensic science, and will also be organized next year.
Friday, October 01, 2010
October 2010
Last month the excellent organized meeting in Moscow of our ENFSI Forensic IT Working group. Many interesting presentations, also from Russian companies in forensic software.
I had also a court case were I had to testify in the week of the meeting, so it was a busy month, since there was also many casework which had to be finished. Also we had an excellent student who developed new and faster algorithms for PRNU comparison. And another student who did very good validation studies for PRNU (camera identification based on Photo Response Non Uniformity).
Also I received a request for posting 20 books you should read. Some of them are very good, however I miss at least a book on digital evidence, since this is becoming more important.
Currently there is also a call for project on forensic investigation of digital evidence from the European Commission in FP7, which attracts several parties.
I am also chairing the organisation of the education days in digital evidence for the Dutch police in November. The program is nearly ready, and it appears there are many new interesting developments.
I had also a court case were I had to testify in the week of the meeting, so it was a busy month, since there was also many casework which had to be finished. Also we had an excellent student who developed new and faster algorithms for PRNU comparison. And another student who did very good validation studies for PRNU (camera identification based on Photo Response Non Uniformity).
Also I received a request for posting 20 books you should read. Some of them are very good, however I miss at least a book on digital evidence, since this is becoming more important.
Currently there is also a call for project on forensic investigation of digital evidence from the European Commission in FP7, which attracts several parties.
I am also chairing the organisation of the education days in digital evidence for the Dutch police in November. The program is nearly ready, and it appears there are many new interesting developments.
Sunday, September 05, 2010
September 2010
This month is again more busy, since everything is back from vacation. This month I will go to the ENFSI Forensic IT Working group meeting in Moscow www.enfsi.eu where I am chairman of. Forensic IT relates to cybercrime, and we see that most cybercrimes crosses borders, so a strong effort from the international community is necessary also to share information an solve the crimes.
Last week I gave a talk to students about cybercrime, and the awareness of students on cybercrime seems not to be high yet. Of course the government and companies can protect us from some of the issues related with cybercrime, however it is also good to have a higher awareness with the users, so cases can be solved more rapidly. There are many efforts also from law enforcements and courts to handle more casework, and often changes of laws are necessary also to cope with it internationally.
Of course we will continue with issues such as observer bias and context effects within forensic science (article Whitman and Koppl). In science it is advised do the cases double blind. In practice some courts will send the case with no additional information to compare, however as they also discuss that the simple fact that the samples are submitted causes a bias.
Last weeks some telephone meetings as member of the Board of the AAFS. Since there is time difference between the Netherlands and the US, it meant that I had to wake up at 2 am in the morning, which I am not used to. However the last meeting was at 11 pm, so that works better. Furthermore also giving some contribution on the Interpol review on image analysis.
Last week I gave a talk to students about cybercrime, and the awareness of students on cybercrime seems not to be high yet. Of course the government and companies can protect us from some of the issues related with cybercrime, however it is also good to have a higher awareness with the users, so cases can be solved more rapidly. There are many efforts also from law enforcements and courts to handle more casework, and often changes of laws are necessary also to cope with it internationally.
Of course we will continue with issues such as observer bias and context effects within forensic science (article Whitman and Koppl). In science it is advised do the cases double blind. In practice some courts will send the case with no additional information to compare, however as they also discuss that the simple fact that the samples are submitted causes a bias.
Last weeks some telephone meetings as member of the Board of the AAFS. Since there is time difference between the Netherlands and the US, it meant that I had to wake up at 2 am in the morning, which I am not used to. However the last meeting was at 11 pm, so that works better. Furthermore also giving some contribution on the Interpol review on image analysis.
Sunday, August 01, 2010
August 2010
August is starting with more casework then usual, and also new cases that come in. It might be caused by a new law on experts in the Netherlands that delays the cases somewhat since all cases need a court order.
Also there is a new policy from Dutch government on blogging for their employees, however as far as I can see the posts that I do fit very well in this policy. The policy is limited to the common practice of blogging.
I have submitted a workshop and a paper for the AAFS before August 1st. Furthermore I am working on reviews for the IWCF International Workshop on Computational Forensics in Japan. Also some reviews for journals, which I have to finish soon.
I am also looking forward to the ENFSI Conference Forensic IT in Moscow , where we are working further on. I think there are many good contributions.
Currently we have some possibilities for students to do a project on Image Analysis and camera Identification. Mostly in R&D I work with students for development and implemenation of new algorithms and also validation projects.
Also there is a new policy from Dutch government on blogging for their employees, however as far as I can see the posts that I do fit very well in this policy. The policy is limited to the common practice of blogging.
I have submitted a workshop and a paper for the AAFS before August 1st. Furthermore I am working on reviews for the IWCF International Workshop on Computational Forensics in Japan. Also some reviews for journals, which I have to finish soon.
I am also looking forward to the ENFSI Conference Forensic IT in Moscow , where we are working further on. I think there are many good contributions.
Currently we have some possibilities for students to do a project on Image Analysis and camera Identification. Mostly in R&D I work with students for development and implemenation of new algorithms and also validation projects.
Tuesday, July 13, 2010
Guest article
When Forensic Science Overreaches
Without a doubt, forensic science has made and continues to make huge contributions to our society. Cases that could never have otherwise been cracked are solved by creative, scientific means. But what happens when forensic science overreaches; when experts make claims that can’t be really be upheld using the strict scientific method? Do innocent people sometimes go to jail simply because your average American citizen has seen too many episodes of CSI and now believes that any statement made by a forensic expert is infallible? The answer to this question is a resounding “yes”.
Forensics as a science.
Like any other scientific field, forensic science is founded on sound scientific principals. Methods of investigation are tested both in the laboratory and the field to determine their accuracy. Statements made in court should be balanced against the limitations of the technology itself and the methods employed to gather the evidence. Unfortunately, juries seem to occasionally have a problem differentiating between deductions made by the investigator and provable facts.
When deduction is confused with fact.
In an article entitled What’s Wrong With Forensic Science that appeared in Newsweek, an example is given of a man who was convicted of a crime he never committed due to the testimony of a forensic scientist who convinced the jury she had matched his designer jeans to evidence at the crime scene. This was in 1989 before the widespread use of DNA evidence. He was acquitted when law enforcement found that his DNA did not match the samples taken at the crime scene. Unfortunately, he wasn’t freed until 2008. The jury had failed to question the deductive process employed by the forensic investigator.
What jurors need to know.
It is paramount for potential jurists to realize that forensic scientists, as humans, can
make mistakes. They can interpret information in the wrong way. They can see connections that aren’t there. A footprint linked to shoes with a certain speck of mud on them should not a guilty verdict make.
Bio: Alexis Bonari is a freelance writer and blog junkie. She is currently a resident blogger at onlinedegrees.org, researching accredited online degree programs. In her spare time, she enjoys square-foot gardening, swimming, and avoiding her laptop.
Without a doubt, forensic science has made and continues to make huge contributions to our society. Cases that could never have otherwise been cracked are solved by creative, scientific means. But what happens when forensic science overreaches; when experts make claims that can’t be really be upheld using the strict scientific method? Do innocent people sometimes go to jail simply because your average American citizen has seen too many episodes of CSI and now believes that any statement made by a forensic expert is infallible? The answer to this question is a resounding “yes”.
Forensics as a science.
Like any other scientific field, forensic science is founded on sound scientific principals. Methods of investigation are tested both in the laboratory and the field to determine their accuracy. Statements made in court should be balanced against the limitations of the technology itself and the methods employed to gather the evidence. Unfortunately, juries seem to occasionally have a problem differentiating between deductions made by the investigator and provable facts.
When deduction is confused with fact.
In an article entitled What’s Wrong With Forensic Science that appeared in Newsweek, an example is given of a man who was convicted of a crime he never committed due to the testimony of a forensic scientist who convinced the jury she had matched his designer jeans to evidence at the crime scene. This was in 1989 before the widespread use of DNA evidence. He was acquitted when law enforcement found that his DNA did not match the samples taken at the crime scene. Unfortunately, he wasn’t freed until 2008. The jury had failed to question the deductive process employed by the forensic investigator.
What jurors need to know.
It is paramount for potential jurists to realize that forensic scientists, as humans, can
make mistakes. They can interpret information in the wrong way. They can see connections that aren’t there. A footprint linked to shoes with a certain speck of mud on them should not a guilty verdict make.
Bio: Alexis Bonari is a freelance writer and blog junkie. She is currently a resident blogger at onlinedegrees.org, researching accredited online degree programs. In her spare time, she enjoys square-foot gardening, swimming, and avoiding her laptop.
Sunday, June 27, 2010
July 2010
Last week I testified in a court case on images, and as usual people become confused with the things they see on CSI and the forensic examination of images and video. I often have to explain that sometimes the image processing which is shown at CSI is not realistic, since we have to cope with the limitation of the sensors, and they have a limited resolution. Often on CCTV it is difficult to do a proper forensic comparison of faces and other objects, and a limited conclusion can be drawn from it. However, there are exceptions.
It is getting summer now, and I do not have any backlog anymore in casework, so we can schedule better, however sometimes cases come in at unexpected moments, and it can change very soon. We are also working on new work for the camera identification, as well as forensic examinations of colours from objects. Before August 1st, I will also propose one or two papers, and perhaps a workshop for next years AAFS www.aafs.org meeting. Furthermore, we are currently working on the ENFSI Forensic IT Working group meeting in Moscow in September www.enfsi.org .
It is getting summer now, and I do not have any backlog anymore in casework, so we can schedule better, however sometimes cases come in at unexpected moments, and it can change very soon. We are also working on new work for the camera identification, as well as forensic examinations of colours from objects. Before August 1st, I will also propose one or two papers, and perhaps a workshop for next years AAFS www.aafs.org meeting. Furthermore, we are currently working on the ENFSI Forensic IT Working group meeting in Moscow in September www.enfsi.org .
Saturday, June 05, 2010
June 2010
Last month was busy with the World Conference on IT where the NFI also signed a MOU with Microsoft. I gave a presentation on the huge efforts that we have to do in reverse engineering of software and hardware for extracting information from devices from manufacturers. Some manufacturers will provide information on their file formats, however others do not know or will not give information, and then we have to spend lots of time on reverse engineering, which also could be spend on examining other products such as reverse engineering malware in forensic casework.
It was also busy with casework and of course the R&D projects. Currently I also have three students that are working on camera identification under the different circumstances. And of course we have many groups of students that are working on various programming tasks, such as our open source software defraser for video analysis and repair. Also there we further progress, as we have version which is much more interactive which will be released soon.
Next week we organize a tour and BBQ as well as a day on forensic video investigation for our customers (currently over 100 people have signed up), and then it gets summer, so somewhat more time to spend on writing reports, articles etc. During the WCIT I was also invited for the Techno Security 2010 in Myrtle Beach to give a presentation, however due to the huge amount of work, I was unable to be there, however I will give some contribution to the conference via Skype.
Friday, May 07, 2010
May 2010
Last month everything went well, and when I was in Moscow I was impressed with the organization of the meeting in September for ENFSI our working group on Forensic IT. This month we will also have the WCIT 2010 in Amsterdam, where I have received the opportunity to speak on Forensic Science and new developments within digital evidence and related areas.
As usual some casework, and currently due to economic situation, also within in forensic science, many labs have issues with budget cuts, since many labs are government owned, and currently many governments have to reduce costs. However as usual, this situation also give opportunities for new developments to have more efficiency and keeping the same, high quality. The last 20 years that I worked, I experienced several of these cuts, however it is not easy to predict what will happen, and each time there are different solutions.
Saturday, April 17, 2010
Fwd: zonsondergang
Monday, April 12, 2010
April 2010
Now we do not have backlogs in casework in image analysis anymore, due to the more structured approach of handling cases. It is also a nice month for preparations. The proficiency test for camera identification is working well, and it is good to see that the different labs have comparable results.
At the end of this month, I will go to the steering committee of ENFSI www.enfsi.org in Moscow, since we will have our meeting from our working group in Moscow in September. Furthermore, I am also working on the security track of www.wcit2010.org in Amsterdam.
And defraser, the software for analysis of video has now a very good preview mode, and it will be available soon via http://sourceforge.net/projects/defraser/ . We are exciting about this option, since it makes the software more interactive.
At the end of this month, I will go to the steering committee of ENFSI www.enfsi.org in Moscow, since we will have our meeting from our working group in Moscow in September. Furthermore, I am also working on the security track of www.wcit2010.org in Amsterdam.
And defraser, the software for analysis of video has now a very good preview mode, and it will be available soon via http://sourceforge.net/projects/defraser/ . We are exciting about this option, since it makes the software more interactive.
Saturday, February 27, 2010
March 2009
The conference of the AAFS in Seattle was excellent. With our completely revised workshop, we had many good comments. Furthermore, our section Digital and Multimedia Sciences also had very good papers, and of course there were many possibilities for exchange and networking. During this meeting I was elected as Director of our Section for the Board of Directors of the AAFS.
Sunday, February 14, 2010
February 2010
A cold month with lots of snow. However also nice developments are going on. The American Academy of Forensic Sciences with the Annual meeting in Seattle will be a very nice event. I will chair the revised workshop on Multimedia Analysis, where we will also have a hands on training on camera identification. Also analysis of video files will be a part of this, and since I am chairman of the Section Digital and Multimedia Sciences, it is nice to see that the field is expanding further. And this week I am also working on many new students that start with projects and developing software for the open source sourceforge.net that we have for forensic software.
We have several students that work in house, however also many have projects that will be run from the universities / hogescholen in the Netherlands themselves. The students Informatics with a minor on Forensics of the Hogeschool Amsterdam will also start with several projects. Sometimes we had extremely good results. Most often this happens if the project is small, well defined, and ofen development in Java or .NET will give good results. Working on existing projects remains a challenge, since often documentation remains an issue.
In the meantime, we are preparing the conference for our Working Group of Forensic IT of ENFSI in Moscow, Russia. It will be in September, more information is available soon. And in the meantime, I have also some casework and other work to do. However good planning is always half of the work.
We have several students that work in house, however also many have projects that will be run from the universities / hogescholen in the Netherlands themselves. The students Informatics with a minor on Forensics of the Hogeschool Amsterdam will also start with several projects. Sometimes we had extremely good results. Most often this happens if the project is small, well defined, and ofen development in Java or .NET will give good results. Working on existing projects remains a challenge, since often documentation remains an issue.
In the meantime, we are preparing the conference for our Working Group of Forensic IT of ENFSI in Moscow, Russia. It will be in September, more information is available soon. And in the meantime, I have also some casework and other work to do. However good planning is always half of the work.
Tuesday, January 19, 2010
January 2010
It seems January gives lots of changes and opportunities. I first had to clean my website from a trojan which appeared only to be detected with one virus scanner. So I cleaned many html-files manually and check all changes in order to have the best effort. I do not think many users have noticed it, however malware is getting more anoying to fix, and it is always a balance between options on a website and freedom on developing and security.
This month I had some interesting discussions also in courses that I gave in digital evidence in a course on Forensic Economics at the University of Leiden. How do we combine evidence, especially in the fields where no numbers are given, however a subjective conclusion. How do we combine conclusions if there is no calibration between the different fields of expertise ? I worked in different fields of expertise and I know that there are differences between experts and between groups of experts and the conclusions that are drawn. The Bayesian rules might help, however still we need calibration.
Last week I went to inaugural lecture of Prof. dr. Ate Kloosterman, Special Chair of Forensic Biology of the Institute for Biodiversity and Ecosystem Dynamics (IBED). It was an excellent overview he gave of the development of DNA analysis of the years. After that there was a reception, and it was very nice to meet also my many of the past colleagues that I did not see for a long time. After a while I was not feeling well, and went out of the reception, and I fell down, since I was very dizzy. I could not stand anymore, since I was sick. In no time there was an ambulance available and checked my heart and I went rapidly to the hospital OLVG nearby. After many examinations in the hospital (MRI, CT-scans, blood, ECG etc) it appeared that I had very probably an infection of the organ of balance, and had to stay in the hospital for two days. For me it was very difficult, since I could not do anything, I could not stand, or send email or send sms messages anymore. I felt very bad that this happened during this party.
However now I am recovering, and total recovery is expected in two weeks (I still use some anti-nausea medicine Primperan), and walking and going on the bicycle should be more easy. It is also a good period to think, so I had to be at home for a week.
Of course also busy with emailing and organizing. The AAFS conference is also soon, so I am preparing. I also have some time to do some reviews, and we are now solliciting for new students for projects. We are also happy that the proficiency test on PRNU is working well, and currently my colleague Wiger and I submitted a new article on a new and faster algorithm for computing PRNU.
This month I had some interesting discussions also in courses that I gave in digital evidence in a course on Forensic Economics at the University of Leiden. How do we combine evidence, especially in the fields where no numbers are given, however a subjective conclusion. How do we combine conclusions if there is no calibration between the different fields of expertise ? I worked in different fields of expertise and I know that there are differences between experts and between groups of experts and the conclusions that are drawn. The Bayesian rules might help, however still we need calibration.
Last week I went to inaugural lecture of Prof. dr. Ate Kloosterman, Special Chair of Forensic Biology of the Institute for Biodiversity and Ecosystem Dynamics (IBED). It was an excellent overview he gave of the development of DNA analysis of the years. After that there was a reception, and it was very nice to meet also my many of the past colleagues that I did not see for a long time. After a while I was not feeling well, and went out of the reception, and I fell down, since I was very dizzy. I could not stand anymore, since I was sick. In no time there was an ambulance available and checked my heart and I went rapidly to the hospital OLVG nearby. After many examinations in the hospital (MRI, CT-scans, blood, ECG etc) it appeared that I had very probably an infection of the organ of balance, and had to stay in the hospital for two days. For me it was very difficult, since I could not do anything, I could not stand, or send email or send sms messages anymore. I felt very bad that this happened during this party.
However now I am recovering, and total recovery is expected in two weeks (I still use some anti-nausea medicine Primperan), and walking and going on the bicycle should be more easy. It is also a good period to think, so I had to be at home for a week.
Of course also busy with emailing and organizing. The AAFS conference is also soon, so I am preparing. I also have some time to do some reviews, and we are now solliciting for new students for projects. We are also happy that the proficiency test on PRNU is working well, and currently my colleague Wiger and I submitted a new article on a new and faster algorithm for computing PRNU.
Subscribe to:
Posts (Atom)